ja kanis je brazier m stevenson nw...

Treatment of established osteoporosis:a systematic review and cost–utilityanalysis

JA KanisJE BrazierM StevensonNW CalvertM Lloyd Jones

HTAHealth Technology Assessment NHS R&D HTA Programme

Health Technology Assessment 2002; Vol. 6: No. 29

Copyright notice

© Queen's Printer and Controller of HMSO 2002 HTA reports may be freely reproduced for the purposes of private research and study and may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising Violations should be reported to [email protected] NOT FOR RESALE. This electronic document is available free of charge from http://www.ncchta.org Applications for commercial reproduction should be addressed to The National Coordinating Centre for Health Technology Assessment, Mailpoint 728, Boldrewood, University of Southampton, Southampton, SO16 7PX, UK

© Queen's Printer and Controller of HMSO

How to obtain copies of this and other HTA Programme reports.An electronic version of this publication, in Adobe Acrobat format, is available for downloading free ofcharge for personal use from the HTA website (http://www.hta.ac.uk). A fully searchable CD-ROM isalso available (see below).

Printed copies of HTA monographs cost £20 each (post and packing free in the UK) to both public andprivate sector purchasers from our Despatch Agents.

Non-UK purchasers will have to pay a small fee for post and packing. For European countries the cost is£2 per monograph and for the rest of the world £3 per monograph.

You can order HTA monographs from our Despatch Agents:

– fax (with credit card or official purchase order) – post (with credit card or official purchase order or cheque)– phone during office hours (credit card only).

Additionally the HTA website allows you either to pay securely by credit card or to print out yourorder and then post or fax it.

Contact details are as follows:HTA Despatch Email: [email protected]/o Direct Mail Works Ltd Tel: 02392 492 0004 Oakwood Business Centre Fax: 02392 478 555Downley, HAVANT PO9 2NP, UK Fax from outside the UK: +44 2392 478 555

NHS libraries can subscribe free of charge. Public libraries can subscribe at a very reduced cost of £100 for each volume (normally comprising 30–40 titles). The commercial subscription rate is £300 per volume. Please see our website for details. Subscriptions can only be purchased for the current orforthcoming volume.

Payment methods

Paying by chequeIf you pay by cheque, the cheque must be in pounds sterling, made payable to Direct Mail Works Ltdand drawn on a bank with a UK address.

Paying by credit cardThe following cards are accepted by phone, fax, post or via the website ordering pages: Delta, Eurocard,Mastercard, Solo, Switch and Visa. We advise against sending credit card details in a plain email.

Paying by official purchase orderYou can post or fax these, but they must be from public bodies (i.e. NHS or universities) within the UK.We cannot at present accept purchase orders from commercial companies or from outside the UK.

How do I get a copy of HTA on CD?

Please use the form on the HTA website (www.hta.ac.uk/htacd.htm). Or contact Direct Mail Works (seecontact details above) by email, post, fax or phone. HTA on CD is currently free of charge worldwide.

The website also provides information about the HTA Programme and lists the membership of the variouscommittees.

HTA

Treatment of established osteoporosis:a systematic review and cost–utilityanalysis

JA Kanis1 *

JE Brazier2

M Stevenson2

NW Calvert2

M Lloyd Jones2

1 WHO Collaborating Centre for Metabolic Bone Diseases,University of Sheffield Medical School, Sheffield, UK

2 Sheffield Centre for Health and Related Research,University of Sheffield, Sheffield, UK

* Corresponding author

Declared competing interests of the authors: Professor John Kanis has workedwith and received funding from many pharmaceutical companies and non-governmentalorganisations that are involved in osteoporosis research.

Published February 2003

This report should be referenced as follows:

Kanis JA, Brazier JE, Stevenson M, Calvert NW, Lloyd Jones M.Treatment of establishedosteoporosis: a systematic review and cost–utility analysis. Health Technol Assess2002;6(29).

Health Technology Assessment is indexed in Index Medicus/MEDLINE and Excerpta Medica/EMBASE. Copies of the Executive Summaries are available from the NCCHTA website(see opposite).

Related publications:Brazier JE, Green C, Kanis JA, on behalf of the Committee of Scientific Advisors, International OsteoporosisFoundation.A systematic review of health state utility values for osteoporosis related conditions. Osteoporos Int2002;13:768–77.

Stevenson M. Gaussian process modelling in conjunction with individual patient simulation modelling.A casestudy describing the calculation of cost-effectiveness ratios for the treatment of osteoporosis. Med Decis Making(in press).

NHS R&D HTA Programme

The NHS R&D Health Technology Assessment (HTA) Programme was set up in 1993 to ensure that high-quality research information on the costs, effectiveness and broader impact of health

technologies is produced in the most efficient way for those who use, manage and provide care in the NHS.

Initially, six HTA panels (pharmaceuticals, acute sector, primary and community care, diagnosticsand imaging, population screening, methodology) helped to set the research priorities for the HTAProgramme. However, during the past few years there have been a number of changes in and aroundNHS R&D, such as the establishment of the National Institute for Clinical Excellence (NICE) andthe creation of three new research programmes: Service Delivery and Organisation (SDO); New and Emerging Applications of Technology (NEAT); and the Methodology Programme.

This has meant that the HTA panels can now focus more explicitly on health technologies (‘health technologies’ are broadly defined to include all interventions used to promote health,prevent and treat disease, and improve rehabilitation and long-term care) rather than settings of care. Therefore the panel structure has been redefined and replaced by three new panels:Pharmaceuticals; Therapeutic Procedures (including devices and operations); and DiagnosticTechnologies and Screening.

The HTA Programme continues to commission both primary and secondary research. The HTACommissioning Board, supported by the National Coordinating Centre for Health TechnologyAssessment (NCCHTA), will consider and advise the Programme Director on the best researchprojects to pursue in order to address the research priorities identified by the three HTA panels.

The research reported in this monograph was funded as project number 95/11/04.

The views expressed in this publication are those of the authors and not necessarily those of theHTA Programme or the Department of Health. The editors wish to emphasise that funding andpublication of this research by the NHS should not be taken as implicit support for anyrecommendations made by the authors.

Criteria for inclusion in the HTA monograph seriesReports are published in the HTA monograph series if (1) they have resulted from workcommissioned for the HTA Programme, and (2) they are of a sufficiently high scientific quality as assessed by the referees and editors.

Reviews in Health Technology Assessment are termed ‘systematic’ when the account of the search,appraisal and synthesis methods (to minimise biases and random errors) would, in theory, permitthe replication of the review by others.

HTA Programme Director: Professor Kent WoodsSeries Editors: Professor Andrew Stevens, Dr Ken Stein, Professor John Gabbay,

Dr Ruairidh Milne and Dr Chris HydeManaging Editors: Sally Bailey and Sarah Llewellyn Lloyd

The editors and publisher have tried to ensure the accuracy of this report but do not accept liabilityfor damages or losses arising from material published in this report. They would like to thank thereferees for their constructive comments on the draft document.

ISSN 1366-5278

© Queen’s Printer and Controller of HMSO 2002

This monograph may be freely reproduced for the purposes of private research and study and may be included in professional journalsprovided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising.

Applications for commercial reproduction should be addressed to The National Coordinating Centre for Health Technology Assessment,Mailpoint 728, Boldrewood, University of Southampton, Southampton, SO16 7PX, UK.

Published by Core Research, Alton, on behalf of the NCCHTA.Printed on acid-free paper in the UK by The Basingstoke Press, Basingstoke. G


© Queen’s Printer and Controller of HMSO 2002. All rights reserved.

List of abbreviations .................................. i

Executive summary .................................... iii

1 Introduction .................................................. 1Established osteoporosis................................ 1Significance of osteoporosis.......................... 1Intervention strategies .................................. 3

2 Therapeutic intervention in osteoporosis .................................................. 7Methodology .................................................. 7Evidence from clinical trials.......................... 10

3 Synthesis of data and discussion .............. 39The quality of evidence ................................ 39The efficacy of intervention.......................... 39Discussion ...................................................... 45

4 Epidemiology, costs and utilities .............. 49Osteoporotic fracture .................................... 49BMD and fracture risk .................................. 52Fracture risk in established osteoporosis .... 54Consequences of fracture.............................. 56Breast cancer and cardiovascular disease .... 58Health state utility values .............................. 59A review of costing ........................................ 67

5 Health economics model .......................... 77Model approach ............................................ 77Overview of model ........................................ 78Population of the model .............................. 80Default state transition probabilities ............ 81Adjustments to the default transitionprobabilities.................................................... 82Treatment ...................................................... 83

6 Results............................................................ 87Analytical approach ...................................... 87Specific treatments ........................................ 88Sensitivity analysis .......................................... 98Clinical vignettes ............................................ 103

7 Discussion and conclusions ........................ 105Treatment effects .......................................... 105Health state utility values in establishedosteoporosis.................................................... 107Hazard functions in established osteoporosis.................................................... 107Constraints of the model .............................. 107Implications for practice .............................. 108Recommendations for further research ...... 108

Acknowledgements .................................... 111

References .................................................... 113Trials meeting the inclusion criteria ............ 125Trials excluded from the systematic review .............................................................. 132

Appendix 1 MEDLINE search strategy for RCTs.......................................................... 135

Appendix 2 Details of handsearching ........ 137

Appendix 3 Quality assessment tool .......... 139

Appendix 4 Details of RCTs (by study) ...... 141

Appendix 5 Summary of intervention studies and quality assessment ...................... 219

Appendix 6 Economic literature search strategy ................................................ 245

Health Technology Assessment reportspublished to date ........................................ 247

Health Technology Assessment Programme .................................................. 253

Contents


i


List of abbreviations

All abbreviations that have been used in this report are listed here unless the abbreviation is well known (e.g. NHS), orit has been used only once, or it is a non-standard abbreviation used only in figures/tables/appendices in which casethe abbreviation is defined in the figure legend or at the end of the table.

BMD bone mineral density

CHD coronary heart disease

CI confidence interval

EQ-5D EuroQol-5 dimension [scale]

FIT Fracture Intervention Trial

GP general practitioner

HCHS Hospital and Community Health Services

HRT hormone replacement therapy

HUI Health Utility Index (-II/-III)

ITT intention-to-treat

LOS length of stay

MI myocardial infarction

NNT number needed to treat [to prevent one fracture]

NOF National Osteoporosis Foundation (USA)

PROOF Prevention of OsteoporoticFractures [study]

QALY quality-adjusted life-year

RCT randomised controlled trial

RR relative risk

SD standard deviation

SERM selective [o]estrogen receptormodulator

SF Short Form (-36 or -6D)

SHEMO Sheffield health economic model of osteoporosis

SG standard gamble

SOF Study of Osteoporotic Fractures

T-score the deviation in units of SD of aBMD value from the mean valuein premenopausal healthy women

TTO time trade-off

Vitamin D cholecalciferol or calciferol

Vitamin D 1-alpha hydroxylated forms of

derivatives vitamin D (calcitriol, alfacalcidol,dihydrotachysterol)

Z-score the deviation in units of SD of aBMD value from the mean valuein individuals of the same age and gender


iii


Background and aimsOsteoporosis is a systemic skeletal disease,characterised by low bone mass and micro-architectural deterioration of bone tissue with a subsequent increase in bone fragility and susceptibility to fracture.

The most serious clinical consequence ofosteoporosis is hip fracture, which increases inincidence exponentially with age and incurs highmorbidity, mortality and healthcare expenditure.Other common fractures occur at the spine,forearm and shoulder.

Osteoporosis is operationally defined by themeasurement of bone mineral density (BMD) at the hip, and is diagnosed in women when BMD is 2.5 standard deviations (SDs) or morebelow the average for young healthy women.Established osteoporosis denotes the disease in the presence of one or more fragility fractures.

A variety of agents are available for the treatmentof osteoporosis. The evidence for their efficacy isexamined and their cost-effectiveness is modelledin established osteoporosis.

Methods

Therapeutic interventionA systematic review was undertaken of all random-ised controlled trials (RCTs) in which fracture was measured as an outcome. RCTs that studiedfracture benefits in patients in whom osteoporosisor osteopaenia was not identified were excluded, as were epidemiological studies, although accountwas taken of these lower levels of evidence in theinterpretation and subsequent analysis of infor-mation. The interventions reviewed were: bisphos-phonates, vitamin D, 1-alpha hydroxylated deriv-atives of vitamin D, calcitonin, calcium, oestrogens,oestrogen-like agents, anabolic steroids, fluoridesalts, thiazide diuretics, raloxifene, vitamin K2,protein supplements and exercise.

Epidemiology, costs and utilitiesThe annual risk of osteoporotic fracture wascharacterised for women from the UK. Fractures

of the hip, spine, distal forearm and humerus weredesignated as being osteoporotic. Collectively, theyaccount for approximately 70% of osteoporoticfractures in postmenopausal women and morethan 70% of the morbidity.

The risk of osteoporotic fractures in women at the threshold for osteoporosis was determinedfrom a published meta-analysis of the relationshipbetween BMD and fracture risk. The risk of such a fracture in the presence of a prior osteoporoticfracture was computed from a published meta-analysis of the relationship between the prioroccurrence of fracture of each type and the risk of a future fracture of each type.

The consequences of fracture on mortality wereassessed for each fracture type. The annual risk of breast cancer, coronary heart disease (CHD)and mortality were reviewed so that extraskeletalrisks and benefits of hormone replacement therapy(HRT) and raloxifene could be modelled.

Costs and utilities were determined for osteo-porosis in the UK by systematic review of the literature.

Health economics modelA model was developed comprising an individualpatient-based approach that simulated whether or not events occurred in each subsequent year for each patient.

Transition states included fracture states (hip,wrist, vertebral and proximal humerus), deathfrom hip fracture, nursing home admission owing to the hip fracture, fatal and non-fatal CHD, fatal and non-fatal breast cancer, and death from other causes.

The model simulated cohorts at fixed ages (50, 60, 70 and 80 years) with established osteoporosis.The proportions of the population with differentfracture types were simulated from the knowndistribution of these fractures at different ages.

Effectiveness was populated from the systematicreview of interventions in osteoporosis. Treatmentswere given for 5 years using a 5-year offset time,except for calcium and calcitonin for which a

Executive summary

Executive summary

iv

3-year offset time was used (in this context, offsettime is the duration for which an effect persistsafter the treatment stops). The analytic frameworkwas set at 10 years. Because of the many uncertain-ties, particularly for hip fracture and extra-skeletalrisks and benefits, extensive sensitivity analyseswere undertaken for each agent.

Results

The results of the systematic review of RCTs indi-cated that bisphosphonates, calcitonin, calcium,fluoride salts and raloxifene reduced the incidenceof vertebral fracture. The bisphosphonate,alendronate, also decreased non-vertebral fracture, including hip fracture.

For several agents, failure to demonstrate efficacy,particularly for hip fracture, was largely due to thelack of appropriate RCTs. Epidemiological evidencesuggested that treatment with calcium, calcitonin,HRT, thiazide diuretics, etidronate and anabolicsteroids decreased hip fracture risk. There was also RCT evidence that calcium plus vitamin Ddecreased fracture risk in patients for whom BMD was not known.

The results for each agent at each age are presentedas a central estimate of cost per quality-adjusted life-year (QALY) gained compared with no treatment.Costs were discounted at 6% and QALYs at 1.5% inbase-case scenarios. The estimate was bounded by a90% confidence interval representing the range of cost–utility that was incurred by 90% of thecombinations of relative risks (RRs) for efficacy.

Cost-effectiveness was graded A–D from the range of cost-effectiveness ratios using a thresholdvalue of £30,000/QALY gained to denote goodcost-effectiveness.

Only those agents that RCT data showed to have significant effectiveness for at least onefracture outcome were tested – raloxifene, HRT,calcium (with and without vitamin D), calcitonin,alendronate, other bisphosphonates, fluoride and alfacalcidol.

It was not cost-effective to treat establishedosteoporosis with raloxifene in the time framemodelled. If cardiovascular benefits were assumed,treatment was only cost-effective compared with no intervention at ages of at least 70 years.

HRT was not cost-effective except below the age of 60 years. However, treatment became cost-

effective from the age of 50 years if the effects on appendicular fractures reported in epidemi-ological studies were included. Additional benefitsfrom reductions in CHD, with additional risks from an increased incidence of breast cancer, did not markedly change the conclusions on cost-effectiveness.

Treatment with calcium alone was cost-effectivecompared with no intervention from age 60 years,assuming an effect only on vertebral fracture risk.Treatment was cost-effective at all ages if effects onappendicular fractures were included, as shown by the RCT data for calcium with vitamin D.

Treatment with calcitonin was not cost-effective atany age largely because of its high costs. Treatmentwith alendronate was only cost-effective from age 70 years onwards.

Since no difference in efficacy between thebisphosphonates could be shown, a pooled analysiswas undertaken using the cost of interventionequivalent to etidronate. ‘Bisphosphonate’ treat-ment was cost-effective from age 60 years solelybecause its therapeutic cost was lower than that for alendronate.

Using the meta-analysis of RCTs, treatment withfluoride was not cost-effective, largely because of a high point estimate for hip fracture risk (RR =1.78). If no adverse effect on hip fracture wasassumed, then treatment became cost-effectivefrom age 60 years.

Compared with no treatment, it was not cost-effective to treat established osteoporosis withalfacalcidol except at ages of 70 years or more.

Further sensitivity analyses were undertaken,focussing on those agents with cost-effectivenessgrades A or B.

Age and cost of intervention were important deter-minants of cost-effectiveness. Cost-effectivenessratios were sensitive to changes in discount ratesfor benefits and in the assumption relating tooffset of effect (offset time). Cost-effectiveness wasmarkedly improved when women with T-scoresunder –2.5 SD were selected.

The results were not markedly affected by the threshold used for cost-effectiveness, poorcompliance, variations in the assumptions aboutmortality after hip fracture, duration of treatmentand duration of analysis. The inclusion of costs foradded years of life had little effect in the elderly


v

but improved cost-effectiveness in women aged up to 60 years. In contrast, the inclusion of allvertebral fractures (in addition to clinically overtfractures) had a marked effect on improving cost-effectiveness.

Conclusions

Cost-effective scenarios for several interventions in the management of established osteoporosiswere identified. Cost-effectiveness ratios decreasewith age. At age 50 years, only HRT and calciumplus vitamin D were cost-effective (assuming thatthe agent would decrease the risk of appendicularfractures at this age). At age 80 years, HRT,calcium with or without vitamin D, alfacalcidol,alendronate and bisphosphonate were all cost-effective.

The conclusions derived are conservative, mainlybecause of the assumptions made in the absence of sufficient data. The conservative assumptionsincluded the following:

(i) not all osteoporotic fractures are included(ii) not all vertebral fractures are included(iii) base-case scenarios are modelled at the

threshold for osteoporosis

(iv) risks of re-fracture in the few years after afracture are likely to be underestimated

(v) vertebral fracture incurs no reversible mortality(vi) long-term effects of osteoporotic fractures

on utilities are ignored.

Thus conclusions that treatments are cost-effective are reasonably secure. In contrast,scenarios shown to be cost-ineffective are lesssecure. As information in these areas becomesavailable, the implications on cost-effectiveness of interventions should be reappraised.

Recommendations for researchIntervention thresholds differ substantially fromdiagnostic thresholds, and should be based on theabsolute fracture probability that depends not onlyon the T-score but also on other independent riskfactors. Health economics assessment based onprobability of fracture is an important area forfurther research.

Other areas for further research arise from gaps in empirical knowledge on utilities and side-effectsthat are amenable to primary research. Furthersecondary research should be undertaken to moreclosely evaluate the impact of vertebral deformities(rather than clinically overt vertebral fractures) on cost-effectiveness.


There is an increasing need for osteoporosismanagement strategies to be placed in an

appropriate health economics perspective. Untilrecently, most economics studies of prevention and treatment of osteoporosis focussed upon theuse of hormone replacement therapy (HRT) at the menopause.1–8 Limited analyses are availablefor the use of non-HRT interventions9–11 but onlyone placed this in a UK setting.11 Moreover, littleor no work has been undertaken to assess the valueof agents on hip fracture – the complication ofosteoporosis that carries the highest consequencefor health. There are now many agents availablefor the treatment of osteoporosis with widelydifferent apparent efficacies and costs. The aim of this study was to examine the cost-effectivenessof treatments available in the UK, specifically inpatients with established osteoporosis.

Established osteoporosis

The internationally agreed definition ofosteoporosis is:

• a systemic skeletal disease characterised by low bone mass and microarchitecturaldeterioration of bone tissue, with a consequentincrease in bone fragility and susceptibility to fracture.12

This definition of osteoporosis captures the notionthat low bone mass is an important component ofthe risk of fracture but that other abnormalitiesoccur in the skeleton, and that non-skeletal factorssuch as falls are also important. Nevertheless, it is only bone mass measured as bone mineral thatcan be presently measured with precision andaccuracy, and this measurement forms the basis for the diagnosis of osteoporosis.

For diagnostic purposes two thresholds of bonemineral density (BMD) have been proposed forCaucasian women based on the T-score.13,14 Thefirst defines the majority of individuals who willsustain a fracture in the future (osteoporosis) and the second, higher threshold may be moreappropriate for investigating the impact of strat-egies to prevent bone loss in women at the meno-pause (low bone mass/osteopaenia). Osteoporosis

is denoted by a BMD value that is at least 2.5standard deviations (SDs) below the young adultmean value (T-score < –2.5 SD). Osteopaenia isdenoted by a T-score that lies between –1 SD and –2.5 SD.

Severe or ‘established’ osteoporosis is as definedabove in the presence of one or more documentedfragility fractures, usually of the wrist, spine or hip.

In a young, healthy population, 15% of women willhave a T-score of less than –1 SD and, thus, haveosteopaenia. Approximately 0.5% will already haveosteoporosis. However, these thresholds apply towomen only and suitable diagnostic cut-off valuesfor men are less secure. It has been suggested thata similar absolute value of BMD to that used forwomen can be taken as the cut-off point for thediagnosis of osteoporosis – namely, a BMD valuethat is 2.5 SD below the average for women.15

Since the introduction of working definitions ofosteoporosis, much attention has focussed on theirapplication in epidemiology, clinical trials andpatient care. Several problems have emerged,however, largely due to the development of newmeasurement techniques applied at many differentsites. It is now clear that the same T-score derivedfrom different sites and techniques yields differentinformation on fracture risk, even when adjust-ments are made for age. Thus, the T-score cannotbe used interchangeably with different techniquesand at different sites. For this reason, the referencestandard adopted in terms of site and technologyfor diagnostic purposes is the hip (femoral neck)using dual energy X-ray absorptiometry.15 Measure-ments at the hip have the highest predictive valuefor hip fracture.16 Moreover, the hip is the site ofgreatest biological relevance since hip fracture isthe dominant complication of osteoporosis interms of morbidity and cost. Hence, in this studythese standards for the definition of osteoporosishave been adopted.

Significance of osteoporosis

The clinical significance of osteoporosis lies in thefractures that arise. Common fractures includevertebral compression fractures, and fractures of


1


Chapter 1

Introduction

Introduction

2

the distal radius and the proximal femur (hipfracture). In addition, when the skeleton isosteoporotic, fractures occur more commonly atmany other sites, including the pelvis, proximalhumerus, distal femur and ribs. Osteoporoticfractures occurring at the spine and the forearmare associated with significant morbidity but themost serious consequences arise in patients withhip fracture, which is associated with a significantincrease in mortality (15–20%), particularly in the elderly.17 Hip fractures account for more than 20% of orthopaedic bed occupancy in theUK, in Scandinavia and in several other coun-tries.13,18 After the age of 45 years, hip fracturesaccount for as high a proportion of hospital bedoccupancy as many other common disorders inwomen, including breast cancer and diabetes.19

The acute hospital costs in the UK for hip, forearm and vertebral fractures in women havebeen estimated at £264 million, and the social care and acute costs for all osteoporotic fractures combined at £727 million.20

The likelihood that any individual will suffer anosteoporotic fracture is considerable. In manyWestern countries, the remaining lifetime prob-ability of a hip fracture in women at the meno-pause lies between 15% and 28%21–23 but variesfrom country to country. In this study, hip fracturerates in Edinburgh have been used (see chapter 4)and, based on these figures, the remaining lifetimerisk at 50 years of age is 14.2% in women and 5.2% in men (Table 1). Even within Europe there is a marked variation in hip fracture probability(see Table 1). For example, men in Sweden have a risk that is comparable to that for women in the UK and significantly higher than that forwomen from Portugal or Turkey. The risk of other common types of osteoporotic fractures isnearly as high,24–27 so that the combined fracture

risk is 30–40%. Thus, more than one-third of adult women will sustain one or more osteo-porotic fractures in their lifetime. This estimate is conservative since it does not include fractures at other sites and only takes into account thosevertebral fractures that come to clinical attention;hence, the true risk of fracture is higher.

These indices of fracture risk compare with thelifetime risk in women aged 50 years of 9–12% for breast cancer and 30–40% for cardiovasculardisease.13 This indicates the widespread prevalenceof osteoporosis in society. In comparison, the risksfor men are about one-third of those in womenand are even lower for forearm fractures; however,they still represent a considerable burden.

The frequency of many osteoporotic fractures,including hip fracture appears to be increasing.The reason for this is two-fold. First, there appearsto have been an increase in age- and gender-specific rates in many countries,28 although there is some evidence that, in England and Wales, the increases in these rates are now stable.29

Second, life expectancy of individuals over the age of 50 years has increased and will continue to do so well into the 22nd century.30 This meansthat, even in the UK, progressively more and more of the population will be elderly and at risk of fracture.31 An approximately four-foldincrease in hip fracture rates between 1990 and2050 on a worldwide basis has been predicted intwo surveys, assuming no increase in the seculartrend.32,33 The increase in fracture frequency isoccurring in both men and women, owing toimprovements in life expectancy. By the year 2025,it has been estimated that there will be as manyfractures in men as there are in women today.Thus, the burden of osteoporosis is set to increase well into the future.

TABLE 1 Lifetime and 10-year probabilities of hip fracture (%) by age and sex in four different areas of Europe

Probability interval Age Malmö Edinburgh Porto Istanbul (years) (years) (Sweden)a (UK)b (Portugal)c (Turkey)c

Men Women Men Women Men Women Men Women

10 50 0.5 0.6 0.4 0.4 0.4 0.4 0.1 0.1

10 60 1.9 2.7 0.7 1.1 0.9 1.3 0.2 0.3

10 70 5.5 10.2 2.2 4.9 1.6 3.2 0.6 0.5

10 80 12.8 24.6 5.7 13.0 2.3 6.0 1.0 1.0

Lifetime 50 13.1 28.5 5.2 14.2 3.2 8.2 1.2 1.2

Hip fracture incidence from different regions taken from:a Kanis, et al., 200024

b Singer, et al., 199825

c Elffors, et al., 199426


3

Intervention strategiesAgainst this background, the development ofintervention strategies in osteoporosis has arousedmuch interest. Two distinct but not mutuallyexclusive strategies can be envisaged. The first is to identify patients at particular risk and to offer an intervention. Examples include theidentification of: women with low BMD, patientswho are likely to fall, and those with certaindiseases or prior fragility fractures. A secondstrategy is a population-based or global strategy in which the aim is to modify a risk factor withinthe general community. For example, if BMD were to be increased by 10% in women, this might decrease the risk of fragility fractures by as much as 50%.24 Such approaches might bedirected at any or all stages of life. Lifestyle factorsthat have been advocated in a global strategyinclude stopping smoking, taking up physicalactivity and improving nutrition in terms ofcalcium and/or vitamin D. There is evidence from randomised controlled studies that inter-vention using a combination of calcium andvitamin D in the elderly living in shelteredaccommodation decreases the risk of hip frac-tures by approximately 28%.35 It is unclear whether or not this is due to the calcium, thevitamin D or the combination. Another success-ful, potentially global strategy is the use of hipprotectors, again in high-risk populations such as nursing home communities. In a recentrandomised study in community healthcarecentres, hip protectors significantly decreased the risk of hip fracture by 60%.36

General global strategiesMore general global strategies such as advice on diet, smoking and exercise are problematic. An example is provided in the case of calciumnutrition. Numerous case–control studies havesuggested that individuals with high intakes ofcalcium have a significantly decreased risk of

hip fracture.9,37 On this basis, there is an impetus,particularly in the USA, to increase the recom-mended dietary allowances for calcium;18 however,the basis of these recommendations is uncertain.So too is their potential impact, which can beillustrated from a large case–control study of hip fractures undertaken in Southern Europe.38

In this study of 3 million individuals, 3000 hipfractures occurred in men and women over the age of 50 years. High intake of calcium in the form of dairy products was associated with asignificant decrease in fracture risk in women(relative risk (RR) = 0.71; 95% confidence interval (CI), 0.58 to 0.87) and a borderline effect in men (RR = 0.81; 95% CI, 0.62 to 1.06).The proportion of the population in whom lowcalcium intakes were observed and who weretherefore at increased risk (attributable risk) was low, so that of these 3000 hip fractures,approximately 110 were associated with a lowcalcium intake. After taking into account whetherrelationships are causal and reversible, and thelikely impact of a public health campaign, thenumbers needed to treat (NNT) to prevent onehip fracture will be large (see Table 2) and, in this example, an NNT value of 2 million was obtained.

These considerations have led to the view thatpopulation-based strategies are not feasible atpresent.18,39,40 The reversible determinants of peak bone mass are conjectural, and no con-trolled studies have been undertaken to modifybone mass in a way that has proved favourable for skeletal health. The global approach of treating all postmenopausal women withoestrogens is unethical and unfeasible at our current state of knowledge.

High-risk strategiesPrevention of osteoporotic fracture is, therefore,more appropriately envisaged today as targetinginterventions at those segments of the community


TABLE 2 An estimate of the effect of a global strategy to increase milk consumption in a population of 3 million men and womenaged 50 years and over37 (Reproduced with permission from Elsevier Science)

Number NNTa

Population, aged 50 years and over 3,000,000

Number of hip fractures 3,000 3,000

Hip fractures associated with low calcium intake 110 27,000– causally associated (80%) 88 34,000– reversible with calcium (70%) 62 48,000

Impact of public health campaign (2.4%) 1.5 2,000,000

a To prevent one hip fracture

Introduction

4

considered to be at high risk. As in the case of a population strategy, this might be directed atappropriate individuals at any age. Since bonemass, at least up to the age of 75 years or so, islargely a function of peak bone mass, it could beargued that high-risk strategies should be directedtowards the optimisation of peak bone mass.Unfortunately, the major contributions to peakbone mass (race and heritability) are immutable,and the impact of other potential factors is ofuncertain value. For this reason, effective high-risk strategies are most appropriately directed inlater life, particularly in women at the menopause,when accelerated bone loss occurs. Interventionsused at this age are largely pharmacological. Ofthese, the greatest attention has been directed at the use of gonadal steroids in women at high risk of fracture.

Population screeningThere are a number of problems that indicate that a widespread approach to the management ofosteoporosis at the menopause with HRT or otherinterventions is problematic. The first is that thetests that might be used in population screeningfor the identification of patients at risk (e.g. BMD)at age 50 years or so lack sensitivity.18 At the hip,the fracture risk increases approximately twofoldfor each SD decrease in BMD. At this gradient of risk, the sensitivity of the test to predict frac-tures over 10 years in women aged 50 years is low. Even if 25% of postmenopausal women were targeted, the sensitivity would remain below50% (Table 3).41 Sensitivity does not increase forthe prediction of all fractures in women aged 65 years. Moreover, the test estimates the short-term fracture risk and, beyond 10 years, riskprediction becomes less secure owing to variablerates of bone loss within the population.42

A second problem is the uncertainty relating to the ultimate impact of intervention for a finite period with HRT (current recommend-ations are 5–10 years) on hip fracture risk muchlater in life.43 Most epidemiological informationindicates that, whereas the risk of hip fracture is markedly reduced in women taking HRT, the effects appear to fall off once treatment is stopped. This offset time appears to occur over a period of 15–20 years.44 Thus, interventionsdirected at the menopause for 5 years are likely to have little or no effect at age 70 or 75 years,when the risk of hip fracture in the communitybegins to increase most markedly. For these andother reasons, no sound case can be made atpresent for screening at the menopause, whereas there may be a case to be made in

later life, when the prevalence of risk factors inaddition to BMD is much higher; however, such an approach has not yet been validated. Until such times, osteoporosis must be tackled using a case-finding strategy.

Case findingThe principles of case finding are to identifyindividuals at risk on the basis of common clinicalrisk factors and, thereafter, to undertake assess-ment of BMD. Whereas the assessment of BMDforms a cornerstone for the diagnosis of osteo-porosis, other risk factors have been identified that contribute to risk independently of BMD.Examples include age, rapid rates of bone loss,prior fragility fractures and a family history offracture. For example, individuals with a priorfragility fracture have a two-fold increase in risk of further fracture, even when this risk is adjustedfor BMD. The combination of a dichotomousvariable (e.g. fracture, present or absent) and acontinuous variable (e.g. BMD) yields a continuousvariable, which in this case has a gradient of riskgreater than that afforded by BMD alone. As canbe seen from Table 3, an increase in the gradient of risk improves sensitivity without trading-offspecificity. Thus, stratification of risk is made more optimal than the use of either of these risk indicators alone. This approach has beenadopted in many practice guidelines.42,45–47

The identification of patients with fragility frac-tures and subsequent measurement of BMD formsan intuitive strategy, provided that intervention can decrease the burden of fracture.

Health economics in osteoporosis is in its infancy.The vast majority of studies in this area have con-centrated on the use of HRT, and only limited dataare available on the use of non-HRT interventions.These non-HRT interventions, however, form thebulk of treatments available for established osteo-porosis. The efficacies of interventions in osteo-porosis and established osteoporosis have nowbeen examined in many published studies. Several bisphosphonates, calcitonin, calcitriol,gonadal steroids and analogues, and calcium have been licensed for use in osteoporosis. Indices of efficacy studied include their effects on BMD and, in some cases, their effects onfracture (reviewed in chapter 2). The logicalelements of a case-finding strategy are thereforecomplete but it requires a justification to competewith other clinical priorities. The most straight-forward scenario is to marshal the arguments in a health economics setting for the treatment of patients with established osteoporosis. Such


5


TAB

LE 3

Es

timat

es o

f pos

itive

pre

dict

ive v

alue

(PP

V),s

ensit

ivity

and

spe

cifici

ty o

f mea

sure

men

ts t

o pr

edict

any

ost

eopo

rotic

frac

ture

ove

r 10

yea

rs o

r to

dea

th in

wom

en a

ged

50 o

r 65

yea

rs,

acco

rdin

g to

diff

eren

t po

pula

tion

cut-o

ffs t

o de

fine

a hi

gh-r

isk c

ateg

ory41

(Rep

rodu

ced

with

per

miss

ion

from

Else

vier

Scie

nce)

Gra

dien

t o

f ri

sk

Hig

h ri

sk c

ateg

ory

(%

of

popu

lati

on)

(RR

/SD

)0.

55

1015

25

PP

VS

ensi

-S

peci

-P

PV

Sen

si-

Spe

ci-

PP

VS

ensi

-S

peci

-P

PV

Sen

si-

Spe

ci-

PP

VS

ensi

-S

peci

-(%

)ti

vity

ficit

y(%

)ti

vity

ficit

y(%

)ti

vity

ficit

y(%

)ti

vity

ficit

y(%

)ti

vity

ficit

y(%

)(%

)(%

)(%

)(%

)(%

)(%

)(%

)(%

)(%

)

Wom

en a

ged

50 y

ears

1.5

17.3

1.5

99.6

12.7

10.7

95.4

10.7

18.0

90.5

10.3

26.1

85.7

8.9

37.6

75.8

2.0

31.3

2.6

99.6

19.2

16.2

95.7

15.1

25.5

91.0

13.8

35.1

86.3

11.3

47.5

76.4

2.5

45.0

3.8

99.7

24.8

21.0

96.0

18.7

31.6

91.4

16.5

41.9

86.7

12.9

54.5

76.9

3.0

56.7

4.8

99.8

29.4

24.9

96.3

21.5

36.3

91.7

18.5

47.0

87.0

14.0

59.2

77.1

4.0

73.0

6.2

99.9

35.8

30.3

96.6

25.1

42.3

92.0

20.9

52.9

87.4

15.1

63.8

77.4

5.0

82.3

7.0

99.9

39.5

33.4

96.8

26.8

45.3

92.2

21.8

55.2

87.5

15.3

64.7

77.5

6.0

87.8

7.4

99.9

41.5

35.0

96.9

27.5

46.5

92.3

21.9

55.6

87.6

15.1

63.7

77.4

Wom

en a

ged

over

65

year

s1.

537

.81.

399

.628

.810

.295

.924

.417

.391

.223

.725

.386

.720

.736

.776

.9

2.0

60.2

2.1

99.8

40.8

14.5

96.6

32.9

23.4

92.2

30.7

32.7

87.9

25.3

45.0

78.3

2.5

76.3

2.7

99.9

49.8

17.7

97.1

39.0

27.7

92.9

35.4

37.7

88.7

28.2

50.1

79.1

3.0

86.1

3.1

99.9

56.2

19.9

97.4

43.1

30.6

93.4

38.3

40.8

89.2

29.8

52.9

79.6

4.0

95.3

3.4

100.

063

.522

.697

.947

.533

.893

.941

.243

.889

.730

.954

.879

.9

5.0

98.7

3.5

100.

067

.023

.898

.149

.234

.994

.141

.744

.489

.830

.454

.079

.8

6.0

100.

23.

610

0.0

68.5

24.3

98.2

49.4

35.1

94.1

41.2

43.8

89.7

29.4

52.1

79.4

Introduction

6

arguments have not yet been fully formalised inthe UK, in part because of the lack of availabilityuntil recently of treatments of proven efficacy.

The increasing awareness of osteoporosis com-bined with the current availability and develop-ment of specific treatments is likely to increase the demand for management of patients withosteoporosis. In the past, management has beenlargely confined to specialists but the increasingavailability of diagnostic tools and well-proventreatments, and the increasing numbers of patients identified, indicate that the burden of management will fall increasingly on theprimary care physician. The aim of this study is to provide a framework for the cost-effective

management of established osteoporosis. Theterms of reference do not cover the treatment of patients with osteoporosis but without fragilityfractures. However, the ultimate goal of effectivetreatment strategies is to prevent the first frac-ture, and the approach used was designed so that these strategies could be considered at a later date.

The evidence for efficacy of interventions inosteoporosis is considered first using an evidence-based approach. In subsequent chapters, theadditional assumptions on risks, costs and qualityof life are described. These data are finallyincorporated into a health economics model to address the major aims of this study.


7

In approaching a systematic review of trials ofefficacy for application to health economics

models, there are two strategies. The first is toreview the randomised controlled trials (RCTs)that examine fracture outcomes. This has theadvantage of incorporating outcomes of clinicalsignificance. The major disadvantage is the relative paucity of trials in which fracture isexamined as the primary outcome compared with studies on BMD. This is partly becauseregulatory authorities worldwide accept studies of BMD as the criteria for efficacy for prevention of osteoporosis.48–50 The second option is to review studies of prevention of bone loss and to infer antifracture efficacy from the knownrelationship between BMD and fracture. Thisapproach was used in early pharmacoeconomicevaluations.8,51 Although the relationship between BMD and fracture risk is well establishedin untreated cohorts, the relationship between a change in BMD and change in fracture risk is less secure.47,50,52 Indeed, recent RCT dataindicated that treatment-induced changes in BMD may underestimate antifracture efficacy –that is, the decrease in fracture rate is greater than can be explained on the basis of themeasurement of BMD alone.53

The approach used was the more direct approach – namely to examine and model fracture outcomes based on RCT evidence from a systematic review of the literature. In view of the paucity of trials that reported fracture as aprimary outcome, studies in which fracture wasreported as a secondary outcome measure werealso included. The advantage of reviewing onlyRCTs is that they provide the highest level ofevidence. A disadvantage is that evidence fromepidemiological studies is not considered. This was appropriate in the context of this study but the censoring of epidemiological evidence isproblematic in the field of osteoporosis. Hipfracture is the most serious consequence ofosteoporosis but very few RCTs on hip fractureoutcomes are available. This is because regulatoryauthorities and, hence, the pharmaceuticalindustry recognise the efficacy of treatments forosteoporosis on the basis of effects on vertebralfracture.48,49 Vertebral fractures are more commonin middle age when hip fractures are rare; hence,

trials in vertebral fracture are more economic toset up in terms of the numbers of patients requiredto demonstrate efficacy. The argument runs that,since osteoporosis is a sys-temic disease andtreatments induce systemic effects, an agent thatdecreases vertebral fracture risk will also do so atother sites vulnerable to osteoporosis.

The epidemiological information is, in general, in keeping with this view. Examples include the use of hormone replacement, calcium, calcitoninand etidronate, for which RCT data indicate avertebral fracture efficacy that is supported bysimilar findings in case–control studies on hipfracture.54,55 There are, however, dangers in making assumptions. The magnitude of hiddenbiases cannot be assessed. Moreover, the magni-tude of effect on hip fracture risk cannot beassumed to be similar to the effect on vertebralfracture. Indeed, as reviewed later, recent studieswith raloxifene have shown a clinically significant(50%) effect of treatment on vertebral fracture risk but an insignificant magnitude of effect onfractures of the appendicular skeleton. For thesereasons, the systemic review and primary analyseswere confined to an RCT information base but other evidence is pointed out and used when appropriate.

Methodology

A systematic review was undertaken to compare the effectiveness of pharmacological and non-pharmacological interventions in preventingosteoporotic fractures in patients with osteopaenia,osteoporosis or established osteoporosis.

Inclusion criteria• Types of participants: only those studies were

included in which the participants had beendiagnosed as having established osteoporosis,osteoporosis or osteopaenia, whether primary or secondary.

• Types of intervention: studies were included inwhich any of the following types of interventionwere used:– bisphosphonates– vitamin D


Chapter 2

Therapeutic intervention in osteoporosis


8

– 1-alpha hydroxylated derivatives of vitamin D(referred to as vitamin D derivatives)

– calcitonin– pharmacological doses of calcium– oestrogens (opposed and unopposed)– oestrogen-like molecules– anabolic steroids– fluoride salts– thiazide diuretics– selective [o]estrogen receptor modulators

(SERMs)– non-pharmacological interventions.

• Outcome measures: all studies were included in which vertebral or non-vertebral fracture was reported.

• Study design: only RCTs were included. Trials were accepted as RCTs if the allocation of patients to treatment groups was described as randomised.

Exclusion criteriaIt had originally been intended to include all rele-vant trials, whatever the language of publication.However, it was only possible to include thosepublished in English, French, German, Italian orSpanish. This led to the exclusion of two studiesthat were published only in Japanese: one was anopen-label randomised trial of vitamin K2,

56 and theother a crossover study comparing alfacalcidol withcalcium – the first year of which was, in effect, anRCT.57 A third study, published only in Japanese,provided a relatively lengthy abstract in English,58

and the information which this offered was used.The report of a further trial was published in bothDanish59 and English,60 and thus was included.

Only published studies were included, with a cut-off date of March 2000.

Literature searchLiterature searches of the electronic databaseslisted in Table 4 were undertaken. Each databasewas searched as far back as possible, with nolanguage restrictions. Update searches were carried out on MEDLINE, EMBASE and theCochrane Library in March 2000.

Search strategies of relevant clinical keywords were developed through reference to publishedstrategies and consultation with experts, and byiterative searching, whereby keywords identified in references retrieved by initial scoping searcheswere used to extend the search strategy and soincrease the sensitivity of retrieval. The strategydeveloped by the Cochrane Collaboration toidentify RCTs was used to retrieve studies relating to the treatment of osteoporosis.

Sensitivity of the keyword strategies was tested bycomparing the results of searches with referencelists from existing reviews.

The search strategy for MEDLINE is presented inappendix 1. The search strategies used for otherdatabases are available from the authors.

The reference lists of relevant studies identifiedthrough the electronic searches were checked.Citation searches on the same references werecarried out using Science Citation Index. Referencelists of published reviews were also checked.Attempts to identify further studies were made byconsulting experts, health technology assessmentand guideline producing agencies, and researchand trials registers via the Internet. Relevantpharmaceutical companies were also invited toprovide up-to-date literature relating to theirproducts. In addition, the five journals identifiedby the electronic searches that yielded the greatestnumbers of relevant articles were handsearchedfrom January 1990 onwards (see appendix 2).

Titles and, when available, abstracts of all studiesidentified in the searches were assessed by a singleresearcher for relevance to the review. In cases ofdoubt, the full article was obtained. To ensure thereliability of the selection process, two subject

TABLE 4 Sources searched

Biological Abstracts

Cochrane Controlled Trials Register/Central

Cochrane Database of Systematic Reviews

Current Research in Britain (CRIB)

Current Research Worldwide (CRIW)

EMBASE

HealthSTAR

Index to Theses

Index to Scientific and Technical Proceedings (Institute for Scientific Information)

Science Citation Index (Institute for ScientificInformation)

MEDLINE

NHS Database of Abstracts of Reviews of Effectiveness(DARE)

NHS HTA

NHS Economic Evaluation Database (NHS EED)

National Research Register (NRR)

Pascal

PubMed

Quality of Life in Medicine bibliography


9

experts each checked a sample of 20 abstractsagainst the inclusion criteria. As a result of thisexercise, it was decided to include only thosestudies in which it was specified that patientssuffered from osteoporosis or osteopaenia. It wasnot sufficient that, on the balance of probability,most or all patients would be likely to haveosteoporosis or osteopaenia because of their age or medical status.

Search resultsElectronic searching yielded a total of 12,378 citations, 118 of which were related tostudies that appeared to fulfil the inclusion criteria for this review. A total of 88 RCTs wereidentified by this means. A further eight relevanttrials were identified from the reference lists ofidentified trials and relevant review articles,61–68

and an additional four by hand-searching only.69–72

No trials were identified by pharmaceuticalcompanies that had not been identified by other means.

Full copies of all apparently eligible studies werethen obtained. A total of 83 individual RCTs metthe review inclusion criteria and are listed after the main reference list (page 125), followed by a list of studies that were excluded, and reasons for this (page 132).

Details of eight trials were only available as con-ference abstracts or publications of comparablelength.70,72–78 Although the methodology of anothertrial79 was mentioned in a longer publication,80

its results were only available from conferenceabstracts.79,81 In addition, as mentioned earlier, it was only possible to use the English-languageabstract of a full Japanese publication.58

Data extractionData from the included studies were extracted by a single reviewer, using a predefined dataextraction form.

Publication biasIn an attempt to assess the magnitude of publi-cation bias, separate funnel plots were drawn,plotting effect size (in terms of the vertebralfracture RR) against sample size, for all inter-ventions for which five or more trials reportedvertebral fracture incidence in terms of thenumbers of patients suffering fractures.82

Meta-analysis of combined dataStudies that met the entry criteria were eligible for inclusion in the meta-analyses, provided thatfracture incidence was reported as the number

of patients sustaining fractures. This enabled thecalculation from published data of the RR ofpatients in the intervention group developing anew fracture or fractures compared with those inthe control group. Studies in which only numbersof fractures or fracture rates (that is, numbers offractures per hundred or thousand patient years)were reported were excluded from the meta-analyses. Their inclusion would have violated thebasic statistical assumption that the occurrence of one event does not increase the likelihood of a subsequent event83 since, once an individualhas suffered an osteoporotic fracture, the risk of a subsequent fracture increases.84,85 In prac-tice, the bias may be small since the number of individuals sustaining multiple vertebralfractures is small.

Ideally, the meta-analyses should have includedonly those studies in which fracture was a primaryendpoint. Sensitivity analyses were thereforeundertaken to assess the effect of excluding trialsin which fracture was not a primary endpoint.

The studies that contributed data to the meta-analyses were also heterogeneous in terms ofpatient characteristics (age, gender, severity ofdisease, etc.), the nature of the interventions used(varieties of drug, doses and methods of adminis-tration), the degree of blinding and standards ofreporting. Since the endpoint of interest wasfracture, it was considered appropriate (paceMeunier, 1999)86 to include open-label studies;however, in relation to vertebral fracture, sensitivityanalyses were carried out to assess the effect ofexcluding those studies in which it was notspecified that the outcome assessors were blinded to treatment status.

The meta-analyses were carried out using thecomputer software package, EasyMA 97b,87

using a fixed-effects model.

Quality assessmentA quality assessment was undertaken to inform the sensitivity analyses.88 The methodologicalquality of all trials that met the inclusion criteriawas assessed using a tool developed by Gillespieand colleagues.89 This was selected because it was intended specifically for the assessment of randomised or quasi-randomised trials ofinterventions designed to prevent fracturesassociated with osteoporosis. It included the following items:

• adequacy of randomisation and masking of randomisation



10

• blinded assessment of outcomes – whether outcome assessors were blind to patients’ treatment allocation

• withdrawals – whether the outcomes of thosewho withdrew were described and included in the analysis

• comparability of groups at baseline• confirmation of diagnosis of hip or other

appendicular skeleton fracture• method of diagnosis of vertebral fracture.

Definitions of the various levels of randomisationand concealment of randomisation derived fromPrendiville and colleagues90 were incorporated inthe tool (see appendix 3).

It is recognised that a quality assessment toolassesses reporting quality and not necessarily the true methodological quality of each study.However, when a trial was reported in more thanone publication, the quality score was calculatedon the basis of the combined data from all relevant publications.

Blinding of the quality assessors to author, institu-tion or journal was considered unnecessary.91,92

One researcher undertook quality assessment. To ensure the reliability of the assessment process,two experts also assessed a sample of five studieseach using the same tool. This exercise resulted in clarification of scoring in relation to the method used to identify vertebral fracture.

Evidence from clinical trials

Each of the eligible studies is summarised inappendix 4. A summary of the studies togetherwith an evaluation of their quality is given inappendix 5, listed by therapeutic class. Thetherapeutic classes included the bisphosphonates,vitamin D and its derivatives, calcitonins, calcium,oestrogens and related compounds, all of whichare licensed for use in osteoporosis in the UK. In addition, fluoride salts, thiazide diuretics,ipriflavone and the anabolic steroids that are used by specialist centres were included. Proteinsupplements were also included, since these havebeen shown to decrease morbidity in hip fracturepatients, together with vitamin K2, which is used in Japan. Finally, physical exercise was reviewed as it is commonly recommended as a lifestyleintervention in osteoporosis.

Studies in which an active intervention wascompared with placebo or no treatment are

discussed first, by intervention, followed by adiscussion of those studies in which two or moreactive interventions were compared. However,evidence relating to side-effects and continuancefrom studies in which active interventions werecompared, has been incorporated into the relevant sections of those studies that used aplacebo or an untreated control group.

BisphosphonatesIn all, 18 RCTs were identified that met theinclusion criteria and in which the effects ofbisphosphonates were compared with those ofplacebo or no treatment in patients with osteo-porosis or osteopaenia.77,93–110 In these studies,patients in both the intervention and controlgroups received calcium and/or vitamin D incomparable doses. Studies in which a bisphos-phonate was compared with another activeintervention are discussed later (page 36).

The studies identified were relatively homo-geneous in that all recruited women with primaryosteoporosis or osteopaenia (details of each studyare summarised in appendix 4). However, thestudies varied in terms of the mean ages ofpatients, their durations, and the drugs and doses used (see appendix 5).

One study was designed as a 2-year, double-blind,randomised, placebo-controlled trial.109 However,after the initial 2 years, patients were allowed tochoose whether to continue the original blindedtreatment or to take calcium alone. Those whocompleted the full 3 years, whether on blindedtherapy or on calcium, were then eligible forinclusion in a 2-year, open-label, follow-up study in which all patients took intermittentcyclical etidronate,100 and were subsequently re-randomised to receive intermittent cyclicaltherapy with either etidronate or placebo.111

Only the results of the original 2-year, double-blind, RCT were used here.

In two studies of risedronate, the arm that was treated with 2.5 mg daily, was discontinuedafter 1 year by an amendment to the trialprotocol.100,106

Three trials were open-label in design.103,104,110

For further details of methodological quality, see appendix 5.

Vertebral fracture was a primary outcome measure in five trials,98,100,106,108,109 whereas inanother trial symptomatic vertebral fracture was a primary outcome measure and


11

morphometric vertebral fracture a secondarymeasure.99 In a further eight studies, vertebralfracture was only included as secondary outcomemeasure,94,96,97,101,103,104,107,110 while in a study that included vertebral fracture as an outcomemeasure, no differentiation between primary andsecondary outcome measures was apparent.95

In two trials, non-vertebral fracture was a primary outcome measure.99,100 Such fractures were a secondary outcome measure in 11 trials.77,94,96–98,101,105,106,108–110

In two studies, symptomatic fractures (vertebraland non-vertebral) were noted only as part of the safety monitoring93,102 and, in one of these,only aggregated figures for vertebral and non-vertebral fracture were reported.93

Vertebral fractures reported in terms of thenumber of patients in each arm suffering suchfractures were reported in 11 trials.94,96–103,106,108,109

In one of these, only symptomatic fractures werereported102 and, in another three, the numbers of patients for whom fracture data were availablewere not specified.94,96,97 In a fourth trial, data were broken down into three periods andcumulative figures for the whole study period were not provided.108

In 12 studies, non-vertebral fractures were reportedin terms of the numbers of patients in each armsustaining such fractures,93,94,97–102,105,106,108–110 and, in one, data were provided on the numbers of patients in each arm suffering hip and wrist fractures.102

The results in terms of vertebral fracture may not have been directly comparable in all cases: in four studies, it was specified that only fracturesin previously unfractured vertebrae wereincluded106,108–110 and, in another, criteria were used that allowed the inclusion of both thesefractures and worsening fractures in previouslyfractured vertebrae – but only fractures in previ-ously normal vertebrae were reported;100 in two studies, it was stated explicitly that furtherfractures in previously fractured vertebrae wereincluded,98,99,107 and this was implicit in a furtherstudy.101 In six studies, it was not clear whether the criteria used would allow the inclusion offurther fractures in already affected verte-brae94–97,103,104 and, in two studies, only sympto-matic vertebral fractures were reported.93,102

The definition of vertebral fracture also variedbetween studies. In two, a minimum reduction

of 15% in vertebral height was required for avertebral deformity to be termed a fracture100,106

and, in a third study, a minimum reduction wasrequired of either 15% in posterior height or 20% in anterior or middle height.104 In eightstudies, a minimum reduction of 20% wasrequired94,98,99,101,103,107–110 and, in another, theresults from two trials were combined, one ofwhich had used a definition of at least 15% andthe other 25%.97 In two studies, the definitionbeing used was not specified.95,96

ResultsInformation was available for alendronate,etidronate, pamidronate and risedronate. At thetime of censoring data collection, no publicationswere available for clodronate. The effects of eachagent for which suitable data were provided areshown in Table 5. The three bisphosphonates hadsignificant effects on vertebral fracture risk, rang-ing from 37% (risedronate) to 57% (etidronate).There was no significant difference in effectbetween treatments.

The effects of bisphosphonates on non-vertebralfractures were less marked than in the case ofvertebral fractures (see Table 5). A significant riskreduction was observed with alendronate andrisedronate but the magnitudes of the risk reduc-tions were 17% and 26%, respectively. Etidronatehad no apparent effect on non-vertebral fractures.In the case of alendronate, the effect on non-vertebral fracture was significantly less than on vertebral fracture. No evidence was found of a significant difference in response between bisphosphonates.

The most extensive data were available foralendronate (Table 6). Treatment was associated


TABLE 5 Efficacy of bisphosphonates on the RR of fracturecompared with a control group

Agent RR (95% CI) p-value

Vertebral fractureAlendronate 0.536 (0.439 to 0.656) < 0.001

Etidronate 0.434 (0.236 to 0.800) 0.013

Risedronate 0.628 (0.506 to 0.779) < 0.001

Pooled 0.569 (0.493 to 0.656) < 0.001

Non-vertebral fractureAlendronate 0.825 (0.736 to 0.926) < 0.001

Etidronate 1.011 (0.681 to 1.501) 0.96

Risedronate 0.737 (0.559 to 0.972) 0.031

Pooled 0.824 (0.745 to 0.931) < 0.001


12

with a significant effect on both vertebral and hip fractures. The effects on forearm and otherfractures were not statistically significant. Efficacyon appendicular fractures appeared to be greaterin patients with prior vertebral fracture. In thenon-fracture arm of the Fracture Intervention Trial (FIT), patients were categorised according to BMD.89 The effects on clinical fractures (in-cluding clinical vertebral fractures) and vertebralfractures assessed morphometrically were greaterin patients with osteoporosis (Table 7) than in those with a T-score > 2.5 SD. There was, however,no significant difference. The RRs in patients with osteoporosis were comparable to our meta-

analysis and, hence, these data were not used in any sensitivity analyses.

In order to characterise effects more thoroughly,the available data for alendronate, etidronate andrisedronate were pooled (Table 8 and Figures 1–5).The pooled data suggested that bisphosphonatesdecreased the risk of vertebral fracture by 43%.The results are all fairly homogeneous (Figure 1).In one study, in which intermittent cyclicaletidronate was used both alone and in combi-nation with phosphorus,109 the combinationappeared more effective than etidronate alone.The pooled data also suggested that bisphos-phonates decreased the risk of non-vertebralosteoporotic fractures by 18% (Table 8 and Figure 2). The effect on vertebral fracture was significantly greater than on non-vertebral fracture.

Non-vertebral fractures were disaggregatedwhenever possible according to fracture site.Bisphosphonates had a significant effect on hip fracture risk – which was reduced by 33%(Table 8). In contrast, the effects on forearm and other non-vertebral sites was smaller, and the 95% CIs exceeded unity (Figures 3–5).

As already mentioned, the criteria for the diagnosisof incident vertebral fractures varied betweenstudies. There were no differences in apparentefficacy according to the criteria used, althoughthe mid-point estimate with the more stringentcriteria was associated with somewhat greaterefficacy (Table 8). This is not surprising since lessstringent criteria capture more false-positives,which decrease the apparent efficacy.112

Very little difference was observed when resultswere pooled from studies with higher quality scores (Table 9). Results were pooled from only

TABLE 6 Efficacy of alendronate on the RR of fracturecompared with a control group

Site of fracture RR (95% CI)

All patientsVertebrala 0.544 (0.448 to 0.659)

Hip 0.611 (0.392 to 0.951)

Wrist 0.866 (0.672 to 1.115)

Other 0.862 (0.740 to 1.003)

All non-vertebral 0.825 (0.736 to 0.926)

Patients with prior fractureVertebral 0.529 (0.408 to 0.687)

Hip 0.497 (0.244 to 1.013)

Wrist 0.528 (0.317 to 0.879)

Other 0.993 (0.763 to 1.293)


Patients without prior fractureVertebral 0.558 (0.387 to 0.805)

Hip 0.795 (0.438 to 1.443)

Wrist 1.188 (0.869 to 1.624)

Other 0.803 (0.662 to 0.967)


a All studies included in the meta-analysis used a 20%criterion for vertebral fracture

TABLE 7 Effects of alendronate on the RR of fracture accordingto T-score

T-score RR (95% CI)

Vertebral fracture (morphometrically assessed)< –2.5 0.50 (0.31 to 0.82)–2.5 to –2.0 0.54 (0.28 to 1.04)–2.0 to 1.6 0.82 (0.33 to 2.07)

Clinical fracture (includes clinical vertebral fractures)< –2.5 0.64 (0.50 to 0.82)–2.5 to –2.0 1.03 (0.77 to 1.39)–2.0 to 1.6 0.82 (0.33 to 1.60)

TABLE 8 Efficacy of bisphosphonates (pooled data) on the RRof fracture

Site of fracture RR (95% CI) p-value

Non-vertebral fracture by siteHip 0.672 (0.459 to 0.983) 0.041

Forearm 0.833 (0.659 to 1.054) 0.13

Other 0.862 (0.741 to 1.003) 0.055

All non-vertebral 0.824 (0.745 to 0.913) < 0.001

Definition of vertebral fracture15% deformity 0.628 (0.506 to 0.779) < 0.001

20% deformity 0.526 (0.435 to 0.637) < 0.001

Combined 0.571 (0.495 to 0.659) < 0.01


13

those studies that scored at or above the meanquality score for all studies of bisphosphonates,that had the appropriate fracture as a primaryendpoint, or whose outcome assessors were statedto be blinded to treatment allocation. When onlythe results from the higher-quality studies wereconsidered, the effects of bisphosphonates inreducing the risk of hip fracture were no longersignificant (see Table 9), an effect caused by thesmaller sample size rather than by any change in point estimate.

The presence of vertebral fractures at studyrecruitment appeared to have an important effect.Vertebral fracture efficacy was similar in patientswith or without prior vertebral fractures (Table 10).In contrast, the bisphosphonates decreased the risk of forearm fractures significantly – by 43% in patients with prevalent fractures – but did not reduce the risk in patients without priorvertebral fractures. The point estimate of the effect on hip fracture was also more marked inpatients with prevalent vertebral fractures than in those without (38% vs 27%), although thedifference was not significant.

Of those studies which presented results relating to vertebral fracture incidence in a form that

could not be used in the pooled estimates, onefound the RR in the aggregated treatment groups to be 0.55 compared with a control group.95 In six, no statistically significant differ-ences were found between treatment and controlgroups,94,97,104,107,108,110 although in three of these it was suggested that there was a trend towards a lower incidence in the treatment groups97,107,110

and, in another, there was a statement thatalthough the difference between groups was not significant over the whole length of the 150-week trial, it became significant in weeks 60–150.108

In several studies, the results of non-vertebralfracture risk were presented in a form that couldnot be used in the pooled estimates. These resultsindicated that the incidence of non-vertebralfracture did not differ between treatment andcontrol groups.77,93,96,102

Side-effectsThe principal side-effects of oral bisphosphonatesare gastrointestinal upsets. In the case of amino-bisphosphonates, these are upper gastrointestinaleffects. For the non-aminobisphosphonates(clodronate and etidronate), the side-effects are confined to the lower gastrointestinal tract.


0 1 2 3 4

RR

Study* Size Quality RR RR(n) (95% CI) (95% CI)

Montessori, 1997 40 83 0.077 (0.001 to 4.392)

Watts, 1990 106 78 0.609 (0.28 to 1.325)

Watts, 1990a 106 78 0.349 (0.134 to 0.912)

Reginster, 2000 409 78 0.6 (0.442 to 0.814)

Liberman, 1995 440 67 0.523 (0.283 to 0.968)

Harris, 1999 819 100 0.659 (0.486 to 0.894)

FIT, 1996b 1014 100 0.53 (0.409 to 0.687)

FIT, 1998c 2216 100 0.558 (0.387 to 0.805)

Total 5150 0.569 (0.493 to 0.656)

FIGURE 1 Vertebral fracture in 5150 patients treated with bisphosphonates compared with controls by mean size of treatment arm(s):RRs with 95% CIs (note linear scale)* As listed on page 125a Phosphate group; b fracture arm; c non-fracture arm


14

In the studies included in this review, alendronatewas associated with adverse upper gastrointestinalevents, which in some instances may have beenassociated with failure to take the drug withadequate quantities of water, or to remain uprightafterwards, or both.113 However, although a highfrequency of adverse upper gastrointestinal eventswas noted in the RCTs of alendronate includedhere, it was not significantly higher in patientstreated with alendronate than in those treated with placebo. It may be relevant that these studiesexcluded patients with gastrointestinal disease, which may account for the difference betweenRCT evidence and clinical practice.

In one of the etidronate trials reviewed here, nostatistically significant differences in adverse effects

were found between the treatment and controlgroups that might have been associated withetidronate (abdominal pain, diarrhoea andnausea),109,114 although the use of phosphate as an activating agent was associated with asubstantially higher reported rate of diarrhoeathan in patients receiving placebo.109 Another study found that 35% of women treated withetidronate without HRT complained of nausea but neither those treated with etidronate with HRT nor those on HRT alone complained of this.110

In a double-blind, placebo-controlled study ofpamidronate, gastrointestinal adverse effects werefound to be equally common in the pamidronateand placebo groups.107 An open-label study, in

0.1 0.3 1 3 10

RR


Wimalawansa, 1998 18 78 1.057 (0.097 to 11.56)

Wimalawansa, 1998a 18 78 0.949 (0.086 to 10.424)

Storm, 1990 33 78 0.84 (0.292 to 2.418)

Montessori, 1997 40 83 1 (0.004 to 251.56)

Clemmesen, 1997b 44 72 1 (0.278 to 3.593)

Clemmesen, 1997c 44 72 2.176 (0.746 to 6.352)

Bone, 1997d 90 78 0.993 (0.527 to 1.872)

Bone, 1997e 90 78 0.582 (0.274 to 1.236)

Bone, 1997f 90 78 0.557 (0.262 to 1.184)

Watts, 1990 316 78 1.234 (0.681 to 2.236)

Watts, 1990g 316 78 0.852 (0.441 to 1.647)

Reginster, 2000 409 78 0.707 (0.473 to 1.058)

Liberman, 1995 440 67 0.787 (0.522 to 1.188)

Harris, 1999 819 100 0.636 (0.416 to 0.971)

Pols, 1999 954 73 0.521 (0.303 to 0.897)

FIT, 1996h 1014 100 0.811 (0.649 to 1.014)

FIT, 1998i 2216 100 0.889 (0.761 to 1.04)

Total 6951 0.824 (0.745 to 0.931)

FIGURE 2 Non-vertebral fracture in 6971 patients treated with bisphosphonates compared with controls by mean size of treatmentarm(s): RRs with 95% CIs (note logarithmic scale)* As listed on page 125a HRT group; b continuous treatment; c intermittent treatment; d 1 mg dose; e 2.5 mg dose; f 5 mg dose; g phosphate group; h fracture arm; i non-fracture arm


15


0.01 0.1 1 10 100

RR


Montessori, 1997 40 83 1 (0.004 to 251.56)

Lindsay, 1999 214 56 1 (0.004 to 255.08)

Liberman, 1995 440 67 0.256 (0.033 to 2.006)

Harris, 1999 819 100 0.802 (0.381 to 1.692)

FIT, 1996a 1014 100 0.497 (0.244 to 1.013)

FIT, 1998b 2216 100 0.795 (0.438 to 1.443)

Total 4743 0.672 (0.459 to 0.983)

FIGURE 3 Hip fracture in 4743 patients treated with bisphosphonates compared with controls by mean size of treatment arm(s): RRswith 95% CIs (note logarithmic scale)* As listed on page 125a Fracture arm; b non-fracture arm

0.01 0.1 1 10 100

RR


Storm, 1990 33 78 1 (0.004 to 250.66)

Montessori, 1997 40 83 1 (0.004 to 251.56)

Lindsay, 1999 214 56 1 (0.084 to 11.852)

Liberman, 1995 440 67 0.338 (0.148 to 0.773)

Harris, 1999 819 100 0.64 (0.331 to 1.235)

FIT, 1996a 1014 100 0.53 (0.319 to 0.882)

FIT, 1998b 2216 100 1.187 (0.869 to 1.622)

Total 4776 0.833 (0.659 to 1.054)

FIGURE 4 Wrist fracture in 4776 patients treated with bisphosphonates compared with controls by mean size of treatment arm(s):RRs with 95% CIs (note logarithmic scale)* As listed on page 125a Fracture arm; b non-fracture arm


16

which intravenous pamidronate was compared with fluoride (see page 36), found that about 30% of patients treated with pamidronate suffereda transient fever. Fever did not affect patientstreated with fluoride but, in this study, gastricintolerance was limited to those treated with fluoride.115

In three of the four studies of risedronate, thedistribution of adverse upper gastrointestinalevents was comparable in the intervention andplacebo groups;97,100,106 in the fourth study, noinformation on adverse events was given.77

ContinuanceIn the studies reviewed here, the percentage of patients receiving bisphosphonates whocompleted the protocol ranged from 58% to95%.94,102 In only four studies was compliancespecifically discussed in terms of both the numberof patients who continued to take the medicationand the proportion of medication that they hadtaken. In the intervention arms of the FIT study,89% of surviving patients in the fracture trial and 81% in the non-fracture trial were still takingthe study medication at the final visit; in bothinstances, 96% of those who continued to take

0.01 0.1 1 10 100

RR


Storm, 1990 33 78 1 (0.004 to 250.66)

FIT, 1996a 1014 100 0.993 (0.763 to 1.293)

FIT, 1998b 2216 100 0.803 (0.667 to 0.967)

Total 3263 0.862 (0.741 to 1.003)

FIGURE 5 Other non-vertebral fractures in 3262 patients treated with bisphosphonates compared with controls by mean size oftreatment arm(s): RRs with 95% CIs (note logarithmic scale)* As listed on page 125a Fracture arm; b non-fracture arm

TABLE 9 Efficacy of bisphosphonates (pooled data) on the RRof fracture according to quality of study


High-quality studiesa

Vertebra 0.574 (0.496 to –0.665) < 0.01

Hip 0.694 (0.470 to 1.023) 0.065

Forearm 0.901 (0.704 to 1.152) 0.41

Other non-vertebral 0.862 (0.741 to 1.003) 0.055

All non-vertebral 0.833 (0.747 to 0.928) < 0.001

Fracture as the primary end-pointVertebra 0.575 (0.490 to 0.675) < 0.001

Non-vertebral 0.854 (0.738 to 0.989) 0.035

Assessors blinded to treatmentVertebra 0.561 (0.477 to 0.659) < 0.001

a Studies with a quality score at the mean or higher forbisphosphonates

TABLE 10 Efficacy of bisphosphonates (pooled data) on the RR of fracture according to the presence or absence of priorvertebral fracture

Site of fracture RR (95% CI)

Prior fractureVertebra 0.575 (0.490 to 0.675)

Hip 0.620 (0.368 to 1.042)

Forearm 0.566 (0.377 to 0.848)

Other non-vertebral No data


No prior fractureVertebra 0.558 (0.387 to 0.805)

Hip 0.795 ( 0.438 to 1.443)

Forearm 1.187 (0.869 to 1.622)


All non-vertebral 0.889 (0.761 to 1.039


17

the medication had taken at least 75% of their pills since their last clinic visit.91,99 In anotherstudy,102 over 90% of patients in the interventionarm were at least 90% compliant with the studymedication. In a third study, 86% of patientsoverall took at least 80% of their medication.However, in this trial, only 62% of those in theintervention arm completed the protocol.106

Compliance (as assessed by tablet count) in the intervention group of a fourth study wasreported as 82%.107 In an additional study (seepage 37), in which monthly intravenous injec-tions of pamidronate were compared with oralfluoride, 100% compliance was achieved in thepamidronate arm.115

A survey of compliance in 813 women treated with alendronate found that while 28.7% statedthat they had discontinued treatment, prescriptionrenewal records suggested that 30.2% had actuallydiscontinued treatment. Gastrointestinal problemswere the most common reason given for dis-continuation – cited by 51.9% of women who had stopped taking the drug.116

Vitamin DNo RCTs were identified that both met theinclusion criteria and compared the effects ofvitamin D with those of placebo or no treatment in patients with osteoporosis or osteopaenia. Trials comparing vitamin D with another activeintervention are discussed later (page 36).

Vitamin D derivativesVitamin D derivatives are the 1-alpha hydroxy-lated forms of vitamin D (calcitriol, alfacalcidoland dihydrotachysterol). Calcitriol is licensed for use in the UK for the treatment of osteo-porosis, whereas in some other countries the other derivatives are used. Apart from effectivedose, there is no evidence that their mechanism of action differs; hence, they are consideredtogether here.

The literature searches identified nine RCTs that met the inclusion criteria and compared the effects of vitamin D derivatives (calcitriol and alfacalcidol) with those of placebo or notreatment in patients with osteoporosis orosteopaenia.63,64,117–123 These trials included those in which patients in both the intervention andcontrol groups received calcium and/or vitamin Din comparable doses. Those studies in whichvitamin D derivatives were compared with another active intervention are discussed later (page 36). Details of each study aresummarised in appendix 4.

The trials varied in terms of their duration, the populations studied, and the drugs and dosesused. Although most of the populations studiedcomprised women with primary osteoporosis, onestudy population was rheumatic disease patientswith steroid-induced osteoporosis.63

In one paper, the data from two similar double-blind, placebo-controlled trials were combined.118

At the end of the first year, all patients in theplacebo arm crossed over to treatment. Hence,only the results of the first, placebo-controlled year are reported here. Two other studies wereopen-label.64,120 For further details of methodo-logical quality, see appendix 5.

Four trials had vertebral fracture as a primaryoutcome measure.64,118,120,123 In three studies,vertebral fracture was a secondary outcomemeasure63,119,133 and in two, in which vertebralfracture was included as an outcome measure,there was no apparent differentiation betweenprimary and secondary outcome measures.117,121

Non-vertebral fractures were a secondary outcomemeasure in three studies,63,122,123 and were men-tioned in a further study that did not appear todifferentiate between primary and secondaryoutcome measures.121

In only five studies were effects on vertebralfractures reported as the numbers of patients ineach arm sustaining fractures.117,119,121–123 In threetrials, non-vertebral fractures were reported interms of the numbers of patients in each armsustaining such fractures.117,122,123 In a further study, only the aggregated figures for vertebral and non-vertebral fractures were reported.63

Women with primary osteoporosis were enrolled in all the trials. The studies differed in terms of their definition of vertebral fracture. In three, aminimum reduction of 15% in vertebral height was required for a vertebral deformity to be termed a fracture,118,119,122 whereas in four aminimum reduction of 20% was required.64,120,121,123

In two studies the definition used was not specified.63,117

ResultsThe available data were pooled and the RRs of fracture in patients treated with vitamin Dderivatives were compared with a control group. There was no evidence that vitamin Dderivatives decreased the risk of either vertebral or non-vertebral fracture (Table 11 and Figures 6 and 7).



18

There was no statistically significant differencebetween calcitriol and alfacalcidol in terms ofefficacy (Table 12), although the central RR estim-ate for alfacalcidol was lower than for calcitriol, for both vertebral and non-vertebral fractures.

Those studies in which a 20% decrease in vertebral height was used to define vertebralfracture appeared to favour treatment more than those in which a 15% definition was used;however, the effect on vertebral fracture risk was not statistically significant using the 20%criterion (Table 13).

Little difference was seen when the analysis wasconfined to the higher-quality studies (see Table 11).

Again, little difference in efficacy was seen in those trials in which only patients with previousfractures were included (see Table 11). No trialswere identified that included only those patients without prior fractures.

Of those studies in which results relating to vertebralfracture incidence were presented in a form thatcould not be used in the pooled estimates, it wasreported in one that calcitriol significantly reducedvertebral fracture rates in the treatment group,118

and in another that alfacalcidol alone loweredfracture incidence compared with untreatedpatients, although its efficacy seemed to be increasedby the simultaneous administration of calcitonin.64

In a third study, the occurrence of new vertebral

TABLE 11 Effects of vitamin D derivatives on the RR offracture compared with a control group according to trial qualityand prior vertebral fracture

Site of fracture RR (95% CI) p-value <:

All patientsVertebra 0.982 (0.615 to 1.569) 0.94

Hip 0.346 (0.020 to 5.893 0.46

Wrist 2.718 (0.160 to 46.154 0.49


All non-vertebral 1.353 (0.348 to 5.257) 0.66

Prior vertebral fractureVertebra 1.003 (0.617 to 1.630) 0.99

Hip 0.980 (0.020 to 48.798) 0.99

Wrist 3.901 (0.144 to 100.00) 0.42




Vertebra 1.030 (0.620 to 1.710) 0.91

Hip 0.346 (0.020 to 5.893) 0.46

Wrist 2.718 (0.160 to 46.154) 0.49



Fracture stated as primary end-pointVertebra 0.757 (0.134 to 4.286)

Assessors blinded to treatmentVertebra 0.576 (0.253 to 1.314) 0.19

a Studies that scored at or above the mean for quality forvitamin D derivatives

0 1 2 3 4

RR


Aloia, 1988 17 53 0.75 (0.223 to 2.525)

Gallagher, 1990a 25 73 1.086 (0.529 to 2.233)

Orimo, 1994 40 67 0.32 (0.073 to 1.401)

Ott, 1989 43 67 1.629 (0.645 to 4.109)

Shiraki, 1996 57 72 0.757 (0.134 to 4.286)

Total 182 0.982 (0.615 to 1.569)

FIGURE 6 Other non-vertebral fractures in 182 patients treated with vitamin D derivatives compared with controls by mean size oftreatment arm(s): RRs with 95% CIs (note linear scale)* As listed on page 125


19

fractures was significantly reduced in patients treatedwith alfacalcidol and calcium relative to placebo.120

Alfacalcidol alone was less effective than alfacalcidolplus calcium. Although calcium alone did notappear to be effective, this may be misleading, as the calcium group of patients was more severelyosteoporotic at baseline than the other groups.

A study by Tilyard and colleagues is commonlycited as a pivotal study of efficacy for calcitriol invertebral osteoporosis.124 In this trial, calcium alonewas compared with calcitriol alone; this is thereforeconsidered as a comparator study (see page 36).Fracture rates were expressed as events per person-year and thus not included in this meta-analysis.

There were no data relating to non-vertebralfracture, other than those which contributed

to the pooled estimates. However, in one study ofcalcitriol in rheumatic disease patients with steroid-induced osteoporosis, the data for vertebral andnon-vertebral fractures were pooled.19 There wasno significant difference between the treatmentand control groups.

As the duration of the studies was relatively short,the long-term effects of vitamin D derivatives onfracture frequency are unknown.

Side-effectsIn several studies, calcitriol was found to beassociated with hypercalciuria or hypercalcaemia in all or most of the patients in the interventiongroup.63,117,7119,122 In most of these studies, this was not sufficiently serious to lead to withdrawalfrom the study but, in another study reviewed


0.010.001 0.1 1 10 100

RR


(a)

Orimo, 1994 40 67 1.104 (0.004 to 277.68)

Ott, 1989 43 67 2.03 (0.457 to 9.025)

Shiraki, 1996 57 72 0.076 (0.001 to 4.348)

Total 140 1.353 (0.348 to 5.257)

(b)

Orimo, 1994 40 67 1.104 (0.004 to 277.68)

Ott, 1989 43 67 0.87 (0.003 to 218.4)

Shiraki, 1996 57 72 0.109 (0.002 to 6.698)

Total 140 0.346 (0.02 to 5.893)

(c)

Orimo, 1994 40 67 1.104 (0.004 to 277.68)

Ott, 1989 43 67 7.831 (0.129 to 475.77)

Shiraki, 1996 57 72 0.983 (0.004 to 248.41)

Total 140 2.718 (0.16 to 46.154)

FIGURE 7 RRs with 95% CIs of (a) non-vertebral fracture, (b) hip fracture and (c) forearm fracture in 140 patients treated withvitamin D derivatives compared with controls by mean size of treatment arm(s) (note logarithmic scale)* As listed on page 125


20

later (page 36), two withdrawals from the calcitriolgroup resulted from persistently elevated serumcalcium.124 In one study, it was considered thathypercalciuria could have been avoided by parenteral administration of the drug.117 Noadverse effects of hypercalcaemia on renal functionwere reported.122 In the study reviewed later, 4% ofwomen withdrew from the calcitriol arm of thestudy because of gastrointestinal symptoms.124

ContinuanceIn the studies reviewed here, the percentage ofpatients receiving vitamin D derivatives who com-pleted the protocol ranged from 65% to 91%.122,123

In only one study was compliance specificallydiscussed in terms of the proportion of medicationtaken by study completers – specified to be 97% in both intervention and control groups.121

CalcitoninCalcitonin is a naturally-occurring, 32-amino acidpeptide that has been used for many years in themanagement of osteoporosis. The agent mostwidely used conforms to the structure of salmoncalcitonin, which has a higher potency thancalcitonin from several other species, includingman. Because of its polypeptide nature, calcitonin

has, for many years, been given as a parenteralinjection and is licensed for use on this basis in the UK. In many regions of the world, however, an intranasal formulation is also available, whichpost-dates the original formulations by many years. For this reason, there are no large RCTs in which antifracture frequency using theparenteral formulation has been examined by today’s standards. In the development ofintranasal calcitonin, the major thrust has been to demonstrate equivalence with parenteralcalcitonin in terms of effects on BMD.

A major use of calcitonin has been in themanagement of acute crush fracture syndrome.Following acute vertebral crush fracture, patientsmay be immobilised and suffer pain, and the risk ofrefracture is high. Calcitonin, given for 1–3 monthsfollowing acute vertebral crush fracture, has beenshown in several RCTs to decrease morbidity,improve remobilisation and prevent immobilisationbone loss.125–127 The mechanism for the decrease inbone pain induced by calcitonin is not known forcertain but may involve the release of endogenousendorphins. The use of calcitonin in this manner isbeyond the scope of this review, which focuses uponits long-term use, but a small effect on bone pain isincluded in the quality-adjusted life-year (QALY)estimate (see chapter 3).

A total of 15 RCTs were identified that met theinclusion criteria and in which the effects ofinjected or intranasal calcitonin were comparedwith those of placebo or no treatment in patientswith osteoporosis or osteopaenia.60,61,64,65,93,128–137

These trials include those in which patients in both the intervention and the control groupsreceived comparable doses of calcium and/orvitamin D, and one in which the effect of addingsequential calcitonin to cyclical parathyroidhormone was studied.131 In a further trial, intra-muscular and intranasal calcitonin were com-pared.132 This was a very small, 6-month trialcomparing the effects of the same dose ofcalcitonin (100 U on alternate days) administeredintramuscularly and intranasally to postmeno-pausal women with osteoporosis. None of thepatients from either arm sustained vertebral ornon-vertebral fractures.

Studies in which calcitonin was compared withother active interventions are discussed later (see page 36). Details of all the studies aresummarised in appendix 4.

The reported trials varied in terms of theirduration, the populations studied and the doses

TABLE 12 Effects of vitamin D derivatives on the RR of fracturecompared with a control group according to agent


Vertebral fractureCalcitriol 1.152 (0.688 to 1.928) 0.59

Alfacalcidol 0.459 (0.149 to 1.414) 0.18

HipCalcitriol 0.870 (0.003 to 218.40)

Alfacalcidol 0.249 (0.009 to 6.768)

ForearmCalcitriol 7.831 (0.129 to 475.77)

Alfacalcidol 1.042 (0.021 to 51.993)

All non-vertebral fracturesCalcitriol 2.030 (0.457 to 9.025)

Alfacalcidol 0.193 (0.007 to 5.068) 0.32

TABLE 13 Effects of vitamin D derivatives on the RR ofvertebral fracture compared with a control group according to the criteria used to diagnose incident vertebral fracture

Fracture definition RR (95% CI) p-value

15% 1.266 (0.717 to 2.235) 0.42

20% 0.459 (0.149 to 1.414) 0.18

Not specified 0.750 (0.223 to 2.525)


21

used. Although in the majority, women withprimary osteoporosis or osteopaenia were enrolled,in one men with established osteoporosis wereenrolled, in another men and women with primaryosteoporosis and, in a further study, men andwomen were recruited with steroid-inducedosteoporosis (see appendix 5).

Nine of the 14 trials (all but one of which used injected calcitonin) were open-label,61,64,65,128,

130,135–137 although in four the outcome assessorswere stated to be blinded to treatment allo-cation.61,64,134,135 Only interim results were availablefor the Prevention of Osteoporotic Fractures(PROOF) study.48,133,138 For further details ofmethodological quality, see appendix 5.

In seven trials, vertebral fracture was a primaryoutcome measure.60,64,128,130,133–135 In three, vertebralfracture was a secondary outcome measure129,131,136

and, in another, symptomatic fractures were notedonly as part of the study’s safety monitoring, andvertebral and non-vertebral fractures were notreported separately.93 In three studies that includedvertebral fracture as an outcome measure, thereappeared to be no differentiation between primaryand secondary outcome measures.61,65,137

The results in terms of vertebral fracture were not directly comparable in all cases. In one study it was stated that only fractures in previouslyunfractured vertebrae were included,135 whereas in another three, criteria were used that explicitly allowed the inclusion as fractures of instances of further collapse in already affected vertebrae.61,129,131

The criterion used to define incident vertebralfracture also varied. In five studies, a minimumreduction of 20% in a vertebral height wasrequired,60,64,131,134,135 and was probably required in a further study.129 In another study, a mini-mum reduction of 25% was required.130 Onlysymptomatic fractures were recorded in one trial93

and may have been recorded in another.137 Thedefinitions used in the remaining trials were not clear.61,65,128,133,136

In only one trial was non-vertebral fracture a primary outcome measure;60 such fractures were a secondary outcome measure in a furtherfive studies,131,133–136 and were mentioned in two further studies in which there appeared to be no differentiation between primary andsecondary outcome measures.61,137 Non-vertebralfractures were not mentioned at all in fivestudies.64,65,128–130

Vertebral fracture incidence, in terms of thenumber of patients in each arm suffering suchfractures, was reported in only three studies.60,133,136

Non-vertebral fracture incidence, in terms of thenumber of patients in each arm suffering suchfractures, was reported in six studies.60,78,131,134–136

ResultsThe effects on vertebral and non-vertebral fracture risk are shown in Table 14. For both theintranasal and the parenteral formulation, the RR decreased substantially but with wide CIs. A significant decrease in fracture risk was con-fined to vertebral fracture in patients given theintranasal formulation.

Apart from the effective dose, which varies because of differing bioavailability, there is noknown difference in ultimate effect of the twoformulations. For this reason the data were pooled. The combined data (Table 15) indicatedthat calcitonin reduces vertebral fracture by 40%.The magnitude of the risk reduction was similarfor all non-vertebral fractures combined but theeffect was not significant (Figures 8–11). There wasno decrease in forearm fractures, with a centralestimate close to unity. Thus, although no singlestudy was large enough to demonstrate significantresults, the pooled data suggested that calcitonindecreased the risk of vertebral fracture. With theexception of forearm fractures, calcitonin may also decrease appendicular fractures to a similardegree, but this has not been demonstrated toconventional levels of statistical significance.

These conclusions do not change when the pooled data are taken only from those studies that scored above the mean quality score for allstudies on calcitonin, that had the relevant frac-ture as a primary endpoint, or that stated that the outcome assessors were blinded to studyoutcome (see Table 15).


TABLE 14 Effects of calcitonin on fracture risk according toformulation used

Mode of delivery RR (95% CI) p-value

Vertebral fractureIntranasal 0.611 (0.419 to 0.891) 0.011

Injected 0.077 (0.001 to 4.247)

Non-vertebral fractureIntranasal 0.511 (0.139 to 1.876) 0.31

Injected 0.589 (0.200 to 1.734) 0.34

No trials were identified in which patients withprior fracture were excluded. Little difference inefficacy was seen in trials in which only patientswith prior fracture were included compared withall trials combined (Table 16).

The study in which a 20% definition of vertebralfracture was used appeared to favour treatmentmore than those studies in which no definition wasprovided; however, the result for the latter groupwas not statistically significant (see Table 16).

As the durations of the studies were relativelyshort, the long-term effect of calcitonin on fracture frequency is not known.

Of the studies in which the results relating tovertebral fracture incidence were presented in a form that could not be used in the pooledestimates, in three a significant reduction in thenumbers of new fractures in the treatment groupwas found compared with a control group.61,134,135

In a further four studies, a trend towards a lowerincidence of fractures was found in patients treatedwith calcitonin compared with control groups, butthe numbers were too small to be of any statistical

significance.65,128,130,137 In one study,64 low-doseintermittent calcitonin failed to lower the rate ofvertebral fracture compared with no treatment,although it appeared to augment the effect ofalfacalcidol, and, in another study,111 in whichcyclical parathyroid hormone was compared withand without the addition of sequential calcitonin,more fractures were found in the calcitonin groupthan in the control group; however, this was notstatistically significant. In one study,129 fracture data were not given on the grounds that very few fractures were found in either group.

In three studies, results for non-vertebral fractureincidence were presented in a form that could not be used in the pooled estimates.61,93,137 In all,fewer fractures were found in patients treated with calcitonin than in controls, but this was notstatistically significant in any of the studies.

Side-effectsIn the trials reviewed here, adverse effects werereported from both injected and intranasalcalcitonin; these included hot flushes128,131,136,139

and gastrointestinal complaints such asnausea.61,128,131,136,139 In some cases, intranasalcalcitonin also irritated the nasal mucosa,60

and was associated with rhinitis and minor local nasal or respiratory disorders.129

However, in one trial it was indicated that, inwomen with established postmenopausal osteo-porosis, intranasal calcitonin was associated withsignificant reductions in intensity of pain, limit-ation of action by pain, and analgesic use.139 This is consistent with the findings of another study, inwhich calcitonin, 50–100 IU, injected either intra-muscularly or subcutaneously daily or on alternatedays for an average of 91.6 ± 47 days, resulted inover 90% of patients with osteoporosis (including


22

TABLE 15 Effects of calcitonin on fracture risk compared with acontrol group according to trial quality

Type of fracture RR (95% CI) p-value <:

All dataVertebral 0.600 (0.412 to 0.874) 0.0077

Hip 0.681 (0.145 to 3.198) 0.61

Wrist 0.947 (0.197 to 4.565) 0.95

Other non-vertebral 0.380 (0.074 to 1.941)



Vertebral 0.611 (0.419 to 0.891) 0.011

Hip 0.634 (0.118 to 3.418) 0.60

Wrist 0.714 (0.123 to 4.146) 0.71



Fracture stated as primary outcomeVertebral 0.611 (0.419 to 0.891) 0.011

Hip 0.969 (0.040 to 23.562) 0.98

Wrist 0.969 (0.040 to 23.562) 0.98



Assessors blinded to treatmentVertebral 0.611 (0.419 to 0.891) 0.011

a Analysis confined to studies in which the quality score was ator above the mean for all studies with calcitonin

TABLE 16 Effects of calcitonin on fracture risk according to thepresence or absence of prior vertebral fracture and the criteriafor diagnosis of incident vertebral fracture

Type of fracture RR (95% CI) p-value

Prior vertebral fractureVertebral 0.685 (0.450 to 1.029)

Hip 0.587 (0.089 to 3.737)

Wrist 0.943 (0.183 to 4.864) 0.66



Diagnostic criterion20% 0.308 (0.113 to 0.838) 0.021

Not specified 0.670 (0.447 to 1.004) 0.052


23


0 1 2 3 4

RR


Ringe, 1987 36 50 0.077 (0.001 to 4.274)

Overgaard, 1992a 52 83 0.36 (0.084 to 1.551)

Overgaard, 1992b 52 83 0.037 (0.001 to 1.984)

Overgaard, 1992c 52 83 0.351 (0.081 to 1.515)

PROOF, 2000c 314 73 0.685 (0.456 to 1.029)

Total 506 0.6 (0.412 to 0.874)

FIGURE 8 Vertebral fracture in 506 patients treated with calcitonin compared with controls by mean size of treatment arm(s): RRswith 95% CIs (note linear scale)* As listed on page 125a 50 IU; b 100 IU; c 200 IU

0.010.001 0.1 1 10 100

RR


Ringe, 1987 19 50 0.556 (0.069 to 4.491)

Hodsman, 1997 20 61 0.879 (0.004 to 214.97)

Rico, 1992 30 72 0.477 (0.103 to 2.209)

Rico, 1995 36 72 1 (0.087 to 11.437)

Overgaard, 1992a 52 83 0.556 (0.065 to 4.736)

Overgaard, 1992b 52 83 0.517 (0.061 to 4.417)

Overgaard, 1992c 52 83 0.108 (0.002 to 6.61)

PROOF, 2000b 314 100 0.974 (0.004 to 248.82)

PROOF, 2000c 314 100 0.968 (0.004 to 247.24)

PROOF, 2000d 311 100 0.978 (0.004 to 249.61)

Total 1200 0.53 (0.224 to 1.254)

FIGURE 9 Non-vertebral fracture in 1200 patients treated with calcitonin compared with controls by mean size of treatment arm(s):RRs with 95% CIs (note logarithmic scale)* As listed on page 125a 50 IU; b 100 IU; c 200 IU; d 400 IU


24

0.010.001 0.1 1 10 100

RR


Ringe, 1987 19 50 1 (0.004 to 246.47)

Hodsman, 1997 20 61 0.879 (0.004 to 214.97)

Rico, 1992 30 72 0.501 (0.06 to 4.195)

Overgaard, 1992a 52 83 1 (0.004 to 251.56)

Overgaard, 1992b 52 83 0.931 (0.004 to 234.35)

Overgaard, 1992c 52 83 0.976 (0.004 to 245.55)

Total 225 0.587 (0.089 to 3.737)

FIGURE 10 Hip fracture in 225 patients treated with calcitonin compared with controls by mean size of treatment arm(s): RRs with95% CIs (note logarithmic scale)* As listed on page 125a 50 IU; b 100 IU; c 200 IU

0.010.001 0.1 1 10 100

RR


Ringe, 1987 19 50 1 (0.004 to 246.47)

Rico, 1992 30 72 0.902 (0.08 to 10.196)

Rico, 1995 36 72 1 (0.087 to 11.437)

Overgaard, 1992a 52 83 1 (0.004 to 252.55)

Overgaard, 1992b 52 83 1 (0.004 to 252.55)

Overgaard, 1992c 52 83 1 (0.004 to 252.55)

Total 241 0.947 (0.197 to 4.565)

FIGURE 11 Wrist fracture in 241 patients treated with calcitonin compared with controls by mean size of treatment arm(s): RRs with95% CIs (note logarithmic scale)* As listed on page 125a 50 IU; b 100 IU; c 200 IU


25

post-traumatic osteoporosis) reporting significantlyless pain both at rest and on moving.140

ContinuanceIn the studies reviewed here, the percentage of patients receiving injected calcitonin whocompleted the protocol ranged from 50% to100%.64,132 For intranasal calcitonin, the range was 57–84%.130,132 Compliance was specificallydiscussed in only one study – 15% of patients were reported to be excluded from analysis of the results because of poor compliance.65

CalciumCalcium is the most widely used agent inosteoporosis alone or in combination with othertreatment modalities. Although many RCTs haveexamined the effects of calcium on BMD and bonefracture,37,141 very few satisfied the entry criteria setfor the purposes of this study.

Two RCTs were identified that met the inclusioncriteria and in which the effects of calcium, with or without vitamin D, were compared with theeffects of placebo or no treatment in patients with osteoporosis or osteopaenia. Those studiesthat compared calcium with another activeintervention are discussed later (see page 36).

In one trial, the spine antifracture and bone-sparing efficacy of calcium was investigated inelderly women with low self-chosen calciumintakes, with and without pre-existing vertebralfractures.142 Although participants were notselected on the basis of low BMD, the study wasdesigned to evaluate vertebral fracture in twogroups – women with and without prevalentvertebral fractures on entry. For logistical reasons, it was necessary to randomise patients to treatment without reference to their prevalentfracture status but, when separated into fractureand non-fracture groups for analysis, thesesubgroups were found to be similar in age andcustomary calcium intake. Only the data on those women with pre-existing vertebral fractures are examined here.

The second trial was a very small open-label pilot trial comparing calcium and vitamin D2

with no treatment in postmenopausal women with established osteoporosis.80

For details of the durations of the studies, and the doses used, see appendix 5. Details of methodological quality are also given inappendix 5; the studies are summarised inappendix 4.

Vertebral fracture was a primary outcome measurein the first study.142 It was also an outcome measurein the second study, in which there appeared to beno differentiation between primary and secondaryoutcome measures.80

The results in terms of vertebral fracture may nothave been directly comparable between the twostudies. In one,142 it was implied that the criteriaused allowed the inclusion of fractures in pre-viously fractured vertebrae, while in the second,80

this was not clear. Similarly, in one study thedefinition of vertebral fracture used required areduction in minimum height of 20%,142 in theother the definition used was not specified.80

In neither study was non-vertebral fracture anoutcome measure.

ResultsIn only one study was vertebral fracture reportedin terms of the number of patients suffering suchfractures.142 Fewer patients suffered fracture in the intervention group than in the control group(RR 0.55, 95% CI, 0.33 to 0.93), suggesting thatcalcium is effective in reducing the risk of vertebralfracture in elderly women with low calcium intakesand prior vertebral fractures. Although in thesecond study a higher fracture rate was found inthe calcium group than in the untreated group,this was not statistically significant.80

Side-effectsNo side-effects were reported by either of the trials reviewed here. However, in a trial reportedlater (page 36), 4% of women withdrew from the calcium arm because of gastrointestinalsymptoms.124

ContinuanceIn the only study that provided this information,71% of patients receiving calcium completed theprotocol.124 Another specifically stated that mediancompliance in women with low self-reportedcalcium intake was 64%; this was an overall figure that included groups with and without pre-existing vertebral fractures.142

OestrogensThe major use of oestrogens is in the prevention of postmenopausal symptoms. Oestrogens are,however, widely recommended for the preventionof osteoporosis. Numerous RCTs have shown thatoestrogens prevent bone loss141 and, on this basis,oestrogens are approved for the prevention ofosteoporosis. With the exception of women withhysterectomy, HRT is prescribed with opposed



26

progestogens, usually on an intermittent basis. This results in withdrawal bleeding and, for this reason, few studies exist in which fractureoutcomes are examined in women well past the menopause.

Four RCTs were identified that met the inclusioncriteria and in which the effects of oestrogens were compared with those of placebo or notreatment in patients with osteoporosis orosteopaenia.79,104,110,143,144 These included trials in which patients in both the intervention andcontrol groups received calcium and/or vitamin Din comparable doses. Trials in which oestrogenswere compared with other active interventions arediscussed later (page 36).

The trials reported here varied in terms of theirdurations and the doses used. They were relativelyhomogeneous in terms of their populations (seeappendix 5).

Two trials were open-label.104,110 For further detailsof methodological quality, see appendix 5. Fullerdetails of each study are presented in appendix 4.

In only one trial was vertebral fracture a primaryoutcome measure.144 In two, it was a secondaryoutcome measure104,110 and, in one study thatincluded vertebral fracture as an outcomemeasure, there was no apparent differentiationbetween primary and secondary outcomemeasures.143 Non-vertebral fractures were asecondary outcome measure in one study.110

The results, in terms of vertebral fracture, were not directly comparable in all cases: whereas one study included only fractures in previouslyunfractured vertebrae,110 another used criteriawhich explicitly allowed the inclusion of frac-tures in vertebrae that were already fractured at baseline.143 In the remainder, it was not clear whether the criteria used would allow the inclusion of further fractures in alreadyaffected vertebrae.

The criterion used to define incident vertebralfracture varied also. In one study, a minimumreduction of 15% in anterior, middle or posteriorheight was required,143 in a second a minimumreduction of either 15% in posterior height or20% in anterior or middle height was needed,104

while in a third a minimum reduction of 25% inanterior, middle or posterior height was required,together with a reduction of 15% or more inarea.110 The definition used was not reported in one study.144

ResultsVertebral fractures were reported as the number of patients sustaining such fractures in only onestudy,143 and non-vertebral fractures were reportedin the same way in another.110 The results fromthese studies are shown in Table 17.

There was no evidence that oestrogens decreasedthe risk of non-vertebral osteoporotic fractures.However, the numbers of patients were too small to provide conclusive data. In relation to vertebralfractures,143 there appeared to be a trend foroestrogens to decrease the risk of fracture (RR = 0.583, 95% CI, 0.262 to 1.301).

Of the studies that did not report fractureincidence in terms of the numbers of patientssuffering such fractures, in one the number of new vertebral fractures was almost identical in both the intervention and control groups104

whereas, in another, treatment was associated with a reduction in fractures of the vertebrae and neck of femur.144 In yet another, a statisticallynon-significant trend towards a lower rate of new vertebral fracture was found in the treatment group.110

Side-effectsOestrogens have a number of associatedextraskeletal effects, both beneficial and adverse.Beneficial effects may include reduction of menopausal symptoms such as hot flushes,145

protection against colorectal cancer146 and also, possibly, improved mood and protectionagainst Alzheimer’s disease.147,148 However,oestrogens may also increase the risk of breastcancer, venous thromboembolic events, gall-bladder disease and, unless opposed byprogestogen, endometrial cancer.145,149–152

It is not clear whether they offer protection against coronary heart disease (CHD).153

The assumptions that were used for modelling are reviewed later (page 47).

TABLE 17 Effects of oestrogen and oestrogen-like molecules onfracture risk


OestrogenVertebral 0.583 (0.262 to 1.301)


IpriflavoneVertebral 0.490 (0.186 to 1.294) 0.15



27

In the studies reviewed here, women treated withHRT were reported to suffer from pelvic con-gestion in one,104 while in another, all withdrawalsin women treated with HRT (either alone or incombination with etidronate) were attributed tooestrogen-related adverse events.110

ContinuanceIn the studies reviewed here, the percentage of patients receiving HRT who completed theprotocols ranged from 83%110 to 92%.143 In noneof the studies was compliance discussed specifically.

In other studies, compliance in patients prescribedHRT for osteoporosis has been found to vary, withreports of 36% and 49% compliance after 1 year,and 61% after 6 months–1 year.154–156

Oestrogen-like moleculesOestrogen-like molecules include tibolone andipriflavone. Several RCTs have shown that tibolonereduces the rate of bone loss. It also alleviatespostmenopausal symptoms but no studies offracture outcomes have been reported withtibolone. Ipriflavone is a flavinoid that appears tohave oestrogen-like activity. It is not licensed foruse in the UK, although it is available in someother countries. It has, however, been used byspecialist centres in the UK. Three RCTs wereidentified that met the inclusion criteria; in thesethe effects of ipriflavone were compared with thoseof placebo or no treatment in patients withosteoporosis or osteopaenia.157–159 In these trials,patients in both intervention and control groupsreceived comparable doses of calcium.

These trials were homogeneous in terms of thepopulations studied. They varied in terms of

duration of treatment and the dose of ipriflavoneused (see appendix 5). Details of methodologicalquality are also given in appendix 5, and sum-maries of each study are presented in appendix 4.

Vertebral fracture was a secondary outcomemeasure in all three trials and non-vertebral frac-ture was also an outcome measure in one trial.159

The studies may not have been comparable interms of the criterion used to define incidentvertebral fracture. In two studies no definition was given, while in the third a 20% reduction in vertebral height was used.157–159

In only two studies was vertebral fracture incidencereported in terms of the number of patients in eacharm sustaining such fractures.158,159 Non-vertebralfracture incidence was reported in only one study.159

ResultsThe results from the two studies that providedusable data in relation to vertebral fracture werepooled and the RRs of fracture in individualstreated with ipriflavone compared with a controlgroup are presented in Table 17 and Figure 12. Both of these studies scored above the meanquality score for relevant studies, both had blinded outcome assessors, and in both onlypatients with prior fractures were included.

The combined analysis did not demonstrate thatipriflavone reduced the risk of either vertebral ornon-vertebral fractures.

In the study that presented results relating tovertebral fracture incidence in a form which could not be used in the pooled estimates,


0 2 4 6 8 10 12 14

RR


Passeri, 1992 14 73 0.2 (0.003 to 13.783)

Passeri, 1995 25 83 0.515 (0.19 to 1.397)

Total 39 0.49 (0.186 to 1.294)

FIGURE 12 Vertebral fracture in 39 patients treated with ipriflavone compared with controls by mean size of treatment arm(s): RRswith 95% CIs (note linear scale)* As listed on page 125


28

new fractures occurred more frequently in thecontrol group than in the treatment group.157

Side-effectsIn a review of the safety data available from 60studies of ipriflavone,160 there was no statisticaldifference between the numbers of patients treatedwith ipriflavone and those in the placebo-treatedcontrol groups who suffered adverse reactions(14.5% versus 16.1%) or discontinued treatmentbecause of adverse drug reactions (5.9% versus5.0%). The majority (approximately 80%) ofcomplaints in those treated with ipriflavone andplacebo alike were gastrointestinal in nature(heartburn, vomiting, gastric or abdominal pain,constipation, diarrhoea). However, skin reactions(rash, itching, erythema) and, to a lesser extent,neurological (headache, depression, drowsiness),musculoskeletal (asthenia, fatigue) and cardio-vascular (tachycardia) symptoms were also noted.Laboratory abnormalities included transientchanges in liver and kidney function tests, and in haematological parameters, but no increase was noted in the occurrence of diseases related to such abnormalities.160

In two of the studies reviewed here, gastro-intestinal complaints (mild gastralgia, diarrhoea)were found to be more common in patientstreated with ipriflavone than in controlgroups,157,158 although in the third study suchsymptoms were equally divided between the two groups.159 In one study, treatment wasassociated with a considerable improvement in pain scores and a significant reduction inanalgesic consumption.157

ContinuanceIn the study that provided the fullest infor-mation on withdrawals, 56% of patients receivingipriflavone completed the protocol.159 Whereas this figure is very low, it is comparable to that for the control group (54%). This was the only study in which compliance was specificallydiscussed in terms of the proportion of medicationtaken by patients who completed the study: ofthese, all used at least 75% of the dispensedmedication during the first year of the study and 93% (93%, treatment group; 92%, controlgroup) continued to use at least 75% of themedication for the whole 2-year period.159

Anabolic steroidsMany anabolic steroids have been used in the management of osteoporosis. Those stillcurrently used include stanozolol, which can be given by mouth, and nandrolone, given by

intermittent injection. Neither are currentlylicensed for use in the UK but are used in specialist centres.

Only one RCT was identified that met theinclusion criteria; the effects of anabolic steroidswere compared with those of placebo in patientswith osteoporosis or osteopaenia.161 The agent used was stanozolol. Studies comparing anabolicsteroids with other active interventions arediscussed later (page 36).

Details of the duration of the trial, the populationstudied and the regimen used are presented inappendix 5, together with details of methodo-logical quality. A summary of the study is presented in appendix 4.

Vertebral fracture was a secondary outcomemeasure in this study. However as the number ofpatients suffering such fractures was not reported,it was not possible to determine the RR of sufferingvertebral fracture in the intervention groupcompared with the control group. There was nostatistically significant difference between the twogroups in terms of numbers of fractures.

Side-effectsThe side-effects of stanozolol include derangementof liver function tests and fluid retention.19 Andro-genic side-effects are comparatively rare. Neverthe-less, in the trial included here,161 22% of patientsreceiving stanozolol (and none in the controlgroup) complained of hoarseness, while 30% com-plained of increased facial hair compared with 9%in the control group. Virilisation is the side-effectmost commonly encountered in women treated with nandrolone decanoate, with reported meanincidences of vocal changes in up to 38% of patientsand hirsutism in 24%.162 However, in another study,no signs of virilisation were reported in patientsreceiving this treatment.163 In a third study, voicelowering was reported in 86% of patients receivingnandrolone compared with 12% in the controlgroup, but no instances of hirsutism were re-ported.164 Although vocal changes are generallyconsidered to be irreversible, a study of the effectsof nandrolone on bone mineral content found that,in the majority of cases, these were reversible onprompt discontinuation of treatment.165

In one study, nandrolone improved pain andmobility significantly more than alfacalcidol.166

ContinuanceIn the only study of anabolic steroids that providedseparate information on the number of completers


29

in the different study arms, 91% of patientsreceiving steroids completed the protocol.161 Noinformation was available relating to compliance in terms of both the number of patients who con-tinued to take the medication and the proportionof medication that they had taken.

FluorideVarious formulations of fluoride have been used in the management of osteoporosis since the1930s. The interest has been in the very markedincreases in cancellous bone mass and, with theexception of parathyroid hormone now underclinical development, fluoride is the sole trulyanabolic agent available. Though widely used insome countries, it is not licensed for use in the UK. It is, however, used by specialist centres, and is generally reserved for those with severe vertebraldisease without evidence of marked osteoporosis at appendicular sites.

In all, 11 RCTs were identified that met the inclu-sion criteria; in these the effects of fluorides werecompared with those of placebo or no treatment inpatients with osteoporosis or osteopaenia.79,80,167–175

The trials included those in which patients fromboth the intervention and control arms receivedcalcium and/or vitamin D in comparable doses;those studies in which fluoride was compared withanother active intervention are discussed later(page 36). One trial, in which the comparison was of continuous versus pulsed dosing offluoride,73 is also discussed briefly below.

The trials reported here varied in terms of theirdurations, the populations studied and the dosesused. Although in the majority, the studies were of postmenopausal women with osteoporosis, inone trial men and women with severe osteopaeniawere studied175 and, in another, men with earlyidiopathic osteoporosis.173

Four trials were open-label,79,80,173,174 although in two of these the outcome assessors were statedto be blinded to treatment allocation.173,174 Forfurther details of methodological quality, seeappendix 5. Details of the studies are presented in appendix 4.

Vertebral fracture was a primary outcome measurein six trials,168–172,174 while in a further three trials,vertebral fracture was included only as secondaryoutcome measure.167,173,175 In two trials that in-cluded vertebral fracture as an outcome measure,no differentiation between primary and secondaryoutcome measures was evident.79,80 In six studies,vertebral fracture incidence was reported as the

number of patients in each arm sustainingfractures.79,168,170,171,173,174

Non-vertebral fracture was a primary outcomemeasure in one trial172 and a secondary outcomemeasure in five others.168,169,171,173,174 Non-vertebralfracture incidence was reported in eight studies asthe numbers of patients in each arm sustainingsuch fractures.168–175

The results in terms of vertebral fracture effectsmay not be directly comparable in all cases: in one study it was stated explicitly that incidentfractures included fractures in already fracturedvertebrae,168 and this was implied in four furtherstudies.171–174 However, it was stated in one study,and implied in another, that recurrent fractureswere not allowed.169,175 In one study, separatefigures were given for patients with new andrecurrent fractures and, also, by specifying thenumbers who were fracture-free, those who hadeither new or recurrent fractures.170 In threestudies, it was not clear whether the criteria usedwould allow the inclusion of further fractures inalready affected vertebrae.79,80,167

The criterion used to define new vertebralfractures also varied in these studies. In two, aminimum reduction in vertebral height of 15% was required,168,172 while in five a minimumreduction of 20% was required169–171,173,174 and inanother 25%.175 In three studies the definitionused was not specified.79,80,167

ResultsThe results from those studies providing usabledata were pooled and the RR of fracture inpatients treated with fluoride compared with a control group are presented in Table 18 and Figures 13–16.

The pooled data indicated that fluoride reducedthe risk of vertebral fracture. The results onvertebral fracture effects were relatively homo-geneous but with one marked outlier79 – a smallstudy with poor reporting quality (Figure 13). In a second outlier, a new fracture was defined as a15% reduction in vertebral height.168 When theresults were pooled from only those studies with aquality score above the mean quality score for allfluoride studies, the efficacy of fluoride in relationto vertebral fractures was enhanced (see Table 18).All these studies used a 20% decrease in vertebralheight to define a new fracture.170,171,173,174

In contrast, there was no evidence that fluoridedecreased the risk of non-vertebral fracture, even



30

when incomplete/stress fractures that are a knownside-effect were excluded from the analysis (seeTable 18 and Figures 13–15), and little differencewas seen when results were pooled from only thosestudies which scored at or above the mean qualityscore for all studies involving fluoride.

When only the results from those studies that hadvertebral fracture as a primary endpoint were used,the RR of vertebral fracture was lower than thatobtained from the overall results, but higher thanthat obtained from the higher quality studies only(see Table 18). The lowest RR of all was obtained bypooling the results from those studies in which itwas stated that the outcome assessors were blindedto treatment allocation. However, the RR of non-vertebral fracture was higher in the one study thathad such fractures as a primary endpoint than forall studies.

Fluoride appeared to be more effective in pre-venting new vertebral fractures in patients withoutprevious vertebral fractures than in those with suchfractures (Table 19); however, the former was basedon one study only, of men with early idiopathicosteoporosis.173 There appeared to be a trend to-wards a greater efficacy in terms of non-vertebralfractures in patients without prior fracture but thiswas not statistically significant (see Table 19).

TABLE 18 Effects of fluoride on the risk of fracture comparedwith a control group according to trial quality


All studiesVertebral 0.794 (0.647 to 0.973) 0.026

Hip 1.778 (0.758 to 4.170) 0.18

Wrist 0.766 (0.371 to 1.583) 0.82




Vertebral 0.350 (0.253 to 0.486) < 0.001

Hip 1.799 (0.759 to 4.264) 0.18

Wrist 0.874 (0.271 to 2.818) 0.82



Fracture stated as primary outcomeVertebral 0.347 (0.230 to 0.523) < 0.001


Assessment blind to treatmentVertebral 0.350 (0.253 to 0.487) < 0.001

a Analysis confined to studies in which the quality score was ator above the mean for all studies

0 1 2 3 4

RR


Gutteridge, 1996 25 33 2.333 (1.384 to 3.933)

Ringe, 1998 32 72 0.25 (0.078 to 0.797)

Kleerekoper, 1991 41 50 1.107 (0.819 to 1.496)

Ringe, 1999a 48 78 0.267 (0.138 to 0.517)

Ringe, 1999b 48 78 0.409 (0.242 to 0.692)

Pak, 1995 55 72 0.435 (0.223 to 0.848)

Reginster, 1998 100 89 0.238 (0.052 to 1.088)

Total 349 0.794 (0.647 to 0.973)

FIGURE 13 Vertebral fracture in 349 patients treated with fluoride compared with controls by mean size of treatment arm(s):RRs with 95% CIs (note linear scale)* As listed on page 125a Cyclic; b continuous


31


0.01 0.1 1 10 100

RR


Ringe, 1998 32 72 0.394 (0.121 to 1.284)

Kleerekoper, 1991 41 50 1.513 (0.681 to 3.363)

Sebert, 1995 47 78 1.088 (0.004 to 274.36)

Ringe, 1999a 48 72 0.371 (0.187 to 0.736)

Ringe, 1999b 48 72 0.609 (0.355 to 1.045)

Pak, 1995 55 72 0.549 (0.114 to 2.646)

Meunier, 1998c 89 78 1.184 (0.576 to 2.435)

Meunier, 1998d 89 78 0.775 (0.325 to 1.848)

Meunier, 1998e 89 78 1.667 (0.866 to 3.211)

Reginster, 1998 100 89 1.089 (0.509 to 2.33)

Riggs, 1990 101 89 1.584 (1.006 to 2.494)

Total 739 0.975 (0.782 to 1.217)

FIGURE 14 Non-vertebral fracture in 739 patients treated with fluoride compared with controls by mean size of treatment arm(s): RRswith 95% CIs (note logarithmic scale)* As listed on page 125a Cyclic; b continuous; c 50 mg; d 150 mg; e 200 mg

0.01 0.1 1 10

RR


Pak, 1995 55 72 0.207 (0.003 to 14.961)

Meunier, 1998a 89 78 0.221 (0.004 to 13.679)

Meunier, 1998b 89 78 0.238 (0.004 to 14.671)

Meunier, 1998c 89 78 4.1 (0.836 to 20.11)

Reginster, 1998 100 89 1 (0.085 to 11.755)

Riggs, 1990 101 89 2.231 (0.621 to 8.018)

Total 523 1.778 (0.758 to 4.17)

FIGURE 15 Hip fracture in 523 patients treated with fluoride compared with controls by mean size of treatment arm(s): RRs with95% CIs (note logarithmic scale)* As listed on page 125a 50 mg; b 150 mg; c 200 mg


32

In the studies in which a 20% decrease in vertebralheight was used to define an incident vertebralfracture, greater apparent efficacy was seen than inthose that used a 15% change or did not specify thecriterion used (Table 20); however, the results forboth the latter cases are not statistically significant.

In those studies in which results relating to verte-bral fracture incidence were presented in a formthat could not be used in the pooled estimates,

no statistically significant difference in fractureincidence was seen between patients treated withfluoride and the control group.80,167,169,172,175

In only one study, other than those which contri-buted to the pooled estimate, was any informationgiven on the incidence of non-vertebral fracture – it was indicated that approximately one-sixth of participants in the treatment group sufferedstress fractures compared with none in the control group.79

In a 3-year, double-blind, randomised trial, theeffects were compared of administering the samedose of sodium monofluorophosphate daily, either continuously or for 3 in every 6 months;73

the mean number of new vertebral fracturesincreased equally in both groups.

Side-effectsFluoride has been associated with painful lowerextremity syndrome. In several of the studies

0.01 0.1 1 10 100

RR


Pak, 1995 55 72 5.183 (0.072 to 374.02)

Meunier, 1998a 89 78 1.993 (0.008 to 506.48)

Meunier, 1998b 89 78 2.139 (0.008 to 543.26)

Meunier, 1998c 89 78 2.17 (0.009 to 551.27)

Reginster, 1998 100 89 0.619 (0.16 to 2.402)

Riggs, 1990 101 89 1 (0.004 to 254.16)

Total 523 0.766 (0.371 to 1.583)

FIGURE 16 Wrist fracture in 523 patients treated with fluoride compared with controls by mean size of treatment arm(s): RRs with95% CIs (note logarithmic scale)* As listed on page 125a 50 mg; b 150 mg; c 200 mg

TABLE 19 Effects of fluoride on fracture risk compared with acontrol group according to the presence or absence of priorvertebral fracture(s)

Type of fracture RR (95% CI)

Prior vertebral fractureVertebral 0.686 (0.544 to 0.864)

Hip 1.953 (0.777 to 4.906)

Wrist 0.829 (0.347 to 1.978)



No prior fractureVertebral a 0.250 (0.078 to 0.797)

Hip No data

Wrist No data



a Based on only one study, in men173

TABLE 20 Effects of fluoride on vertebral fracture riskaccording to the criteria used to define incident vertebral fracture

Fracture definition RR (95% CI) p-value

15% 1.107 (0.819 to 1.496)

20% 0.350 (0.253 to 0.486) < 0.001

Not specified 2.333 (1.384 to 3.993)


33

reviewed here, fluoride treatment was associatedwith a significantly increased incidence of lowerextremity pain.169,172–174,176 In one trial, this wasassociated with significantly increased incidences of osteomalacia and microfractures of the lowerextremities168 and, in another, with a statisticallysignificant increase in incomplete fissure fracturescompared with placebo (RR = 13.00; 95% CI, 3.17 to 53.33).172 In two trials,75,115 patients in thefluoride group withdrew from the study because of stress fractures and, in one of these, somepatients also withdrew from the fluoride groupbecause of gastrointestinal symptoms.75

Fluoride has also been associated with an increasein gastrointestinal complaints and several studiesreported such complaints (pain, bloating, nauseaor a change in bowel habit).168,172,176 The incidenceof such complaints decreased when enteric-coatedtablets or capsules were used.168

In one study,72 a significant reduction in back painwas found in women treated with fluoride with orwithout alfacalcidol but not in those treated withalfacalcidol alone.

ContinuanceIn the studies reviewed here, the percentage ofpatients receiving fluoride who completed theprotocol ranged from 50% to 78%.168,173 In onlythree studies was compliance specifically discussedin terms of both the numbers of patients whocontinued to take the medication and the pro-portions of medication that they had taken. In one study, of the patients who took the medication,only 50% of those in the fluoride group took more than 75% of that medication compared with72% in the placebo group.168 In a second study,evaluable patients in both the fluoride and thecontrol groups took, on average, 87% of theirtablets.171 Finally, in a third study, compliance – as assessed by pill count – was 95% in the fluoride group.170

Thiazide diureticsThiazide diuretics are not licensed for use in the management of osteoporosis. Many studieshave shown that thiazides reduce the rates of bone loss and, in case–control studies, their use is associated with a significant decrease in fracturerisk. However, there were no RCTs that met theinclusion criteria.

SERMsSERMs have oestrogen-like activity at skeletal sitesbut a spectrum of activity at extraskeletal sites thatdiffers significantly from oestrogens. Tamoxifen is

widely used in the management and prevention ofbreast cancer recurrence and, in postmenopausalwomen, has been shown to prevent bone loss and decrease fracture frequency.141 Several otherSERMs are being developed for osteoporosis and raloxifene is now available in the UK.

There were no studies with tamoxifen that wereeligible for inclusion. Two RCTs were identified inwhich the effects of raloxifene were compared withthose of placebo or no treatment in patients withosteoporosis or osteopaenia.177,178 In these trials,both the intervention and control groups receivedcomparable doses of calcium and vitamin D.

Both trials were relatively homogeneous in that both studied postmenopausal women withosteoporosis, although in one study all the women,and in the other only some women, had at leastone vertebral fracture at entry.177,178 The trialsvaried in terms of their duration and the doses ofcalcium and vitamin D used (see appendix 5).

Details of methodological quality are given inappendix 5, and summaries of each study arepresented in appendix 4.

Vertebral fracture was a primary outcome measurein one study177 and a secondary outcome measurein the other.178 Non-vertebral fracture was asecondary outcome measure in both studies.

Vertebral fracture incidence was reported as the number of patients in each arm sustainingfractures in both studies but in only one study was non-vertebral fracture incidence reported asthe number of patients in each arm sustainingsuch fractures.177 In the other study, none of the patients had sustained a hip fracture.178

The results in terms of vertebral fracture effectswere not directly comparable between the twostudies. In one, only fractures in previously un-fractured vertebrae were reported177 whereas, inthe other, the criteria used would allow the inclu-sion of fractures in vertebrae that were alreadyfractured at baseline.178 In addition, the definitionof incident fracture used in one study required aminimum reduction of 20% in vertebral height,177

while in the other a minimum reduction of 15%was required.178

ResultsThe results from both trials were pooled and theRRs of vertebral fracture in patients treated withraloxifene compared with controls are presentedin Table 21 and Figure 17.



34

Raloxifene significantly decreased the risk ofvertebral fracture. The results of the study with thehigher quality score were even more favourable(RR = 0.596; 95% CI, 0.516 to 0.688; p < 0.001);177

however, not only did this study differ from theother in quality but it was also substantially larger,it used a 20% definition of vertebral fracturecriteria and the patients enrolled were not asseverely osteoporotic.

Raloxifene appeared to be as effective in pre-venting new vertebral fractures in women withprevious vertebral fractures than in those without(see Table 21).

Raloxifene, 120 mg daily, appeared to be marginallybut not significantly more effective than 60 mgdaily in preventing vertebral fracture (see Table 21).No data were available to allow this comparison tobe made for non-vertebral fractures.

In contrast, raloxifene has not been shown toreduce the risk of non-vertebral fracture (see Table 21 and Figure 18).

In the study in which non-vertebral fracture data was not provided as the number of patients ineach arm sustaining such fractures, there was nosignificant difference between the numbers offractures in each group.178

Side-effects and complianceLike oestrogen, raloxifene has a number ofassociated side-effects, some adverse and some

TABLE 21 Effects of raloxifene on fracture risk compared witha control group according to the presence of prior vertebralfracture at trial entry and dose used


All studiesVertebral 0.661 (0.579 to 0.755) < 0.001

Hip 1.141 (0.663 to 1.966) 0.63

Wrist 0.887 (0.684 to 1.151)



Prior vertebral fractureVertebral 0.674 (0.581 to 0.780) < 0.001

Hip 2.784 (0.095 to 81.961) 0.55

Wrist No data


All non-vertebral No data

No prior vertebral fractureVertebral 0.575 (0.436 to 0.757) < 0.001

All studies, 60 mg doseVertebral 0.715 (0.595 to 0.868) < 0.001

Hip No data

Wrist No data

All non-vertebral No data

All studies, 120 mg doseVertebral 0.606 (0.449 to 0.735) < 0.001


0 1 2

RR


Lufkin, 1998a 48 61 1.221 (0.762 to 1.955)

Lufkin, 1998b 48 61 1.111 (0.684 to 1.804)

Ettinger, 1999a 1032 89 0.65 (0.533 to 0.792)

Ettinger, 1999b 1032 89 0.54 (0.438 to 0.667)

Total 2160 0.661 (0.579 to 0.755)

FIGURE 17 Vertebral fracture in 2160 patients treated with raloxifene compared with controls by mean size of treatment arm(s): RRswith 95% CIs (note linear scale)* As listed on page 125a 60 mg; b 120 mg


35

potentially beneficial. Most seriously, it increasesthe risk of venous thromboembolism approxi-mately three-fold.177,179 In two studies, hot flusheswere also found to increase in a dose-dependentmanner.177,180 In contrast, in two other studies,treatment with raloxifene was reported as not being associated with a higher incidence of hot flushes compared with placebo at doses of either 60 or 120 mg daily.178,181

Raloxifene also lowers fibrinogen levels, and lowers total and low-density lipoprotein chol-esterol without reducing high-density lipoproteincholesterol.178,180–182 However, as yet there are no data available to suggest that it reduces thenumber of cardiovascular events.183 In one of the trials reported here, it was suggested thatraloxifene may also have a preventive role inrelation to breast cancer.177,182,184 In a multi-centre study of women with osteoporosis,182

the risk of invasive breast cancer was decreased by 76% over the 3-year study period (RR = 0.24;95% CI, 0.13 to 0.44). For all breast cancers, RR = 0.35; 95% CI, 0.21 to 0.58. A combinedanalysis of the available data assessed the effect as RR = 0.42; 95% CI, 0.25 to 0.73.184

The long-term effects are unknown and arepresently being investigated in an independenttrial.182 In this study, the available data for patients with osteoporosis were used (RR = 0.35).

Unlike oestrogens, raloxifene does not causeendometrial hyperplasia;178,181 hence, the risk ofendometrial cancer is not increased.182 It has no

adverse effects on mood or cognitive function185

or on breast tenderness.177,180,181

ContinuanceIn the studies reviewed here, the percentage ofpatients receiving raloxifene who completed theprotocol ranged from 78% to over 90%.48,177,178

In the only study in which there were commentson compliance, 92% of patients took more than80% of the study medication and there were nodifferences in compliance between groups.177

Protein supplementsOnly one RCT was identified that met theinclusion criteria, and in which the effects of an oral protein supplement were compared withplacebo in vitamin-D-replete patients with a recenthip fracture.186 For details of duration, populationstudied and regimens, see appendix 5. Details of the methodological quality are also presented in appendix 5 and details of the study aresummarised in appendix 4.

Vertebral fracture was a secondary outcomemeasure in this study. It was reported only as the number of incident fractures and not as thenumber of patients suffering such fractures.

The definition of incident vertebral fracture usedwas a minimum reduction in vertebral height of20%. It was implied that fractures in already frac-tured vertebrae would be included. Fewer fracturesoccurred in patients in the intervention groupthan in the control group but this was notstatistically significant.


0.01 0.1 1 10 100

RR


Lufkin, 1998a 48 61 1.046 (0.004 to 263.53)

Lufkin, 1998b 48 61 5 (0.07 to 359.5)

Ettinger, 1999 1032 89 1.115 (0.642 to 1.934)

Total 1128 1.141 (0.663 to 1.966)

FIGURE 18 Hip fracture in 1128 patients treated with raloxifene compared with controls by mean size of treatment arm(s): RRs with95% CIs (note logarithmic scale)* As listed on page 125a 60 mg; b 120 mg


36

Associated effectsPatients who received protein supplements hadsignificantly shorter stays in rehabilitation hospitalsthan those in the control group.186

ContinuanceOf the patients in the treatment group, 73%completed the study compared with 81% in thecontrol group.186 No information relating tocompliance was given.

Vitamin K2

Vitamin K2 is available for the management ofosteoporosis in Japan but not in the UK. One RCTwas identified that met the inclusion criteria, and inwhich the effects of vitamin K2 were compared withplacebo in ambulatory women with osteoporosis.68

For details of duration, population studied and theregimen, see appendix 5. Details of methodologicalquality are also given in appendix 5, and a summaryof the study is presented in appendix 4.

Vertebral fracture was a primary outcome measurein this study but was reported only in terms of thenumber of incident fractures, rather than the num-ber of patients sustaining fractures.68 The defini-tion of incident vertebral fracture used in thisstudy was a reduction in vertebral height of at least20%. It was implied that fractures in already frac-tured vertebrae would be included. Non-vertebralfracture was a secondary outcome measure.

There were 13 vertebral and one non-vertebralfractures in 77 women in the intervention group,and 30 vertebral and five non-vertebral fractures inthe 64 women in the control group. The combinedincidence of vertebral and non-vertebral fractureswas significantly lower (p = 0.0273) in the vitaminK2 group than in the control group.

Side-effects and continuanceNo information was given regarding either side-effects or continuance.

ExerciseIn addition to the view that exercise is beneficialfor peak skeletal development, exercise is com-monly used in individuals with established osteo-porosis either to decrease the rate of bone loss or to improve confidence and coordination, andthereby decrease the liability of falls.19,141 This raises the question as to whether exercise mightdecrease the risk of fracture. One open-label RCTwas identified that met the inclusion criteria.187

In this the effect was investigated of brisk walkingon BMD, the number of falls and the rate of spinalfractures in postmenopausal women who had

sustained an upper limb fracture in the previous 2 years. For details of duration, population studiedand regimen, see appendix 5. Details of methodo-logical quality are also presented in appendix 5,and the trial is summarised in appendix 4.

Vertebral fracture was a secondary outcomemeasure of this study187 and was reported as thegroup mean total number of fractures per year.The definition used was a 25% difference betweenanterior and posterior vertebral heights; only onefracture per vertebra was counted.

Clinical fractures formed a secondary outcomemeasure; these were reported only as the number ofincident fractures and the rate per 100 person-years.No significant differences were reported between theintervention and control groups in relation to eitherradiographic or clinical fractures. During the courseof the study, six fractures occurred in patients in theintervention group and four in the control group. Itwas not specified how many fractures were clinicallydiagnosed or radiographically diagnosed, althoughtwo women from each group were reported to havesustained fractures following a fall.

Associated effectsBrisk walking has been shown to be beneficial inincreasing high-density lipoproteins and reducingobesity.188,189 However, in the trial reviewed here, itwas also found to be associated with a significantlyincreased risk of falling.187

ContinuanceCompliance with this intervention was poor. Theinitial acceptability of the intervention was low: only 33% of the women who were contacted agreedto take part in the study, and only 19% of thoseinitially contacted completed 2 years. Dropouts weresaid to be evenly distributed between individuals inthe brisk walking group and those in the controlgroup, who undertook upper limb exercises. Rea-sons for withdrawal after randomisation were givenas unwillingness to continue (24%), illness (6%),death (1%), exercise-related trauma (1%) and otherunspecified difficulties (9%). The study was of arelatively poor inner-city population and hence, theauthors suggested, this level of compliance mightnot be typical of other areas.187

Comparisons with active treatmentsA total of 25 RCTs were identified in which one active intervention was compared withanother.58,64,70,72,74–76,79,93,104.109,110,115,124,139,163,164,166,176,

190–195 The trials varied in their duration, thepopulations studied and the regimens used (see appendix 5). Of these trials, 18 were


37

either stated to be open-label or there was no implication that they were blinded, for instance, by reference to the use ofplacebos,64,72,74–76,79,93,104,109,110,115,124,139,164,176,190,191,

193,194 although in six of them the outcome assessors had been blinded to treatmentstatus.64,124,164,190,191,193 For further details of their methodological quality, see appendix 5.

Vertebral fracture was a primary outcome measurein seven studies70,75,109,124,139,166,176 and a secondaryoutcome measure in a further ten.58,72,104,110,115,

190,192–195 In six studies which included vertebralfracture as an outcome measure, there was noapparent differentiation between primary andsecondary outcome measures.74,76,79,163,164,191

Non-vertebral fracture was a primary outcomemeasure in one study193 and a secondary outcomemeasure in a further eight.70,75,109,110,115,124,176,192 It was mentioned in two further studies that did not appear to differentiate between primary and secondary outcome measures.74,191

In one study, data were collected on symptomaticfractures only as part of adverse event reporting;vertebral and non-vertebral fractures were notreported separately.93

In 13 studies, vertebral fracture incidence wasreported as the number of patients in each armsustaining such fractures.58,70,72,75,109,124,163,166,190–193,195

In nine studies, non-vertebral fracture incidencewas reported as the number of patients in eacharm sustaining fractures.79,93,109,110,115,124,176,192,193

In one study,70 which was published only in abstract form, the results were presented withoutunmasking the two groups.

The results from those studies that providedinformation on the numbers of patients sustainingvertebral and non-vertebral fractures, are sum-marised in Tables 22 and 23. In the majority ofstudies, there was no statistically significantdifference between the treatments being com-pared. However, cyclical etidronate, 400 mg


TABLE 22 Comparisons with active treatments: RR of vertebral fracture from all studies providing adequate data

Studya Intervention Comparison RR (95% CI)

Arthur, 1990 Calcitriol + calcium Vitamin D2 + calcium 1.444 (0.006 to 323.97)

Ebeling, 1998 Group Ab Group Bb 8.550 (1.194 to 61.229)

Falch, 1987 Calcitriol Vitamin D3 1.539 (0.649 to 3.650)

Fujita, 1993 Cyclical etidronate, 200 mg/day Alfacalcidol 0.444 (0.192 to 1.027)

Fujita, 1993 Cyclical etidronate, 400 mg/day Alfacalcidol 0.347 (0.142 to 0.847)

Geusens, 1986 Intramuscular nandrolone decanoate Alfacalcidol 0.714 (0.325 to 1.569)

Geusens, 1986 Intramuscular nandrolone decanoate Intravenous calcium infusions 0.682 (0.318 to 1.460)

Geusens, 1986 Alfacalcidol Intravenous calcium infusions + 0.955 (0.524 to 1.739)placebo of alfacalcidol

Guañabens, 1996 Fluoride Cyclical etidronate 0.816 (0.349 to 1.989)

Gutteridge, 1990b HRT + sodium fluoride HRT 0.508 (0.200 to 1.293)

Lems, 1997 Cyclical etidronate, sodium fluoride Cyclical etidronate and calcium 1.826 (0.616 to 5.418)and calcium

Mamelle, 1988 Sodium fluoride, calcium and Other regimens prescribed by 39.2% versus 50.8%vitamin D2 French physicians

Pak, 1989 Cyclical calcitriol, intermittent slow- Intermittent slow-release 2.739 (0.619 to 12.121)release sodium fluoride and calcium sodium fluoride and calcium

Rozhinskaya, 1999 Pooled fluoride + calcium and fluoride, Alfacalcidol 0.909 (0.156 to 5.302)alfacalcidol + calcium groups

Shiraki, 1999 Alendronate + calcium Alfacalcidol + calcium 1.049 (0.152 to 7.219)

Watts, 1990 Etidronate and calcium Sodium-potassium phosphate, 0.938 (0.367 to 2.396)placebo and calcium

Watts, 1990 Etidronate, phosphate and calcium Sodium-potassium phosphate, 0.592 (0.201 to 1.745)placebo and calcium

Watts, 1990 Etidronate, phosphate and calcium Etidronate and calcium 1.584 (0.537 to 4.669)

a See list of trials meeting inclusion criteria (page 125)b Calcitriol or calcium, allocation not specified


38

daily, appeared to be more effective in pre-venting vertebral fracture than alfacalcidol, 1 µg daily,58 and, more tantalisingly, one of the two treatments compared by Ebeling andcolleagues70 was significantly more effective than the other in preventing such fractures;however, as noted above, the two groups were not unblinded. In one study, a significantreduction in vertebral fracture incidence in the second and third years of treatment was found in the calcitriol group compared with the calcium group.124

In relation to vertebral fracture, of those studiesthat did not provide information on the number of patients sustaining vertebral fracture, no signifi-cant difference between treatment groups wasfound in six.74,76,104,110,164,166 In one study, a signifi-cant reduction was found in the rate of vertebralfracture in patients treated with calcitonin

compared with those treated with calcium.139 In a second study, a statistically non-significant trendtowards a lower incidence of vertebral fracture was found in patients treated with pamidronatecompared with those treated with fluoride.115

In a third study, calcitonin appeared to be moreeffective in preventing vertebral fracture than the other treatments used.194 A lower incidence of vertebral deformity was found in a fourth study in patients treated with calcitonin andalfacalcidol compared with those treated witheither calcitonin or alfacalcidol alone, andalfacalcidol alone appeared more effective thancalcitonin alone; however, the numbers were too small to be conclusive.64

In one study, there was no information on thenumber of patients sustaining non-vertebralfracture and no significant difference betweentreatment groups.191

TABLE 23 Comparisons with active treatments: RR of non-vertebral fracture from all studies providing adequate data

Studya Intervention Comparison RR (95% CI)

Adami, 1995 Alendronate, 10 mg/day, + calcium Intranasal calcitonin + calcium 1.102 (0.094 to 12.876)

Adami, 1995 Alendronate, 20 mg/day, + calcium Intranasal calcitonin + calcium 1.041 (0.089 to 12.173)

Lems, 1997 Cyclical etidronate, sodium fluoride Cyclical etidronate and calcium 1.638 (0.642 to 4.180)and calcium

Mamelle, 1988 Sodium fluoride, calcium and Other regimens prescribed by 1.019 (0.570 to 1.824)vitamin D2 French physicians

Pak, 1989 Cyclical calcitriol, intermittent Intermittent slow-release 0.508 (0.062 to 4.181)slow-release sodium fluoride sodium fluoride and calciumand calcium

Thiébaud, 1994 Intravenous pamidronate, calcium Fluoride, 20–30 mg/day, 0.200 (0.003 to 13.888)and vitamin D calcium and vitamin D

Tilyard, 1992 Calcitriol Calcium 0.496 (0.246 to 0.999)

Watts, 1990 Etidronate and calcium Sodium-potassium phosphate, 1.684 (0.873 to 3.248)placebo and calcium

Watts, 1990 Etidronate, sodium-potassium Sodium-potassium phosphate, 1.163 (0.569 to 2.379)phosphate and calcium placebo and calcium

Watts, 1990 Etidronate, sodium-potassium Etidronate and calcium 0.691 (0.371 to 1.287)phosphate and calcium

Wimalawansa, 1998 Etidronate, calcium and vitamin D HRT, calcium and vitamin D 1.057 (0.097 to 11.560)

Wimalawansa, 1998 Etidronate, HRT, calcium and vitamin D HRT, calcium and vitamin D 0.949 (0.086 to 10.424)

Wimalawansa, 1998 Etidronate, HRT, calcium and vitamin D Etidronate, calcium and 0.897 (0.082 to 9.836)vitamin D

a See list of trials meeting inclusion criteria (page 125)


39

In the previous chapter, the evidence for efficacyof a wide range of interventions was reviewed,

based on the literature available from RCTs. The quality of the search strategies used with the electronic databases was such that few trialswere identified by other means, such as by hand-searching or from reference lists. It thereforeseems likely that few published studies have beenmissed. Only two of the identified trials wereexcluded because of language restrictions.

The potential to examine publication bias withineach therapeutic class was limited by the numberof studies available. The asymmetry of the funnelplot relating to bisphosphonates suggests that anumber of small studies with effect sizes distri-buted around the null value may remain un-published. Because of the smaller numbers of relevant studies identified, the situation inrelation to vitamin D derivatives and fluoride is less clear (see appendix 5).

The quality of evidence

As published, most of the trials had potentialmethodological weaknesses. A summary of thequality assessment is provided in Table 24.

The quality of trial reporting appears to haveimproved over the last 20 years, as shown when the studies are categorised by year of mainpublication (Table 25). This improvement becomesmore marked when those studies are removed that are available only as abstracts, rather than as full publications.

There has also been a tendency over time for trialsto become larger. In those studies whose primarypublication predated 1990, the mean number ofpatients was 92 (range 34–466), compared with 118 (range 14–414) in those published between1990 and 1994, and 531 (range 22–7705) in those published in or post-1995.

The efficacy of intervention

As noted above, there was some heterogeneitybetween studies in terms of study populations,


Chapter 3

Synthesis of data and discussion

TABLE 24 Summary of quality assessment according to quality criteria

Included RCTsn (%)

Was randomisation to the study groups blinded?1. States random but no description 66 (80)

or quasi-randomised

2. Small but real chance of disclosure 10 (12)of assignment

3. Method does not allow disclosure 7 (8)of assignment

Were assessors of outcome blinded to treatment status?1. Not mentioned 44 (53)

2. Moderate chance of unblinding 0of assessors

3. Action taken to blind assessors, or 39 (47)outcomes such that bias is unlikely

Were the outcomes of patients who withdrewdescribed and included in the analysis?1. Not mentioned or states number of 15 (18)

withdrawals only

2. States numbers and reasons for 36 (43)withdrawal, but analysis unmodified

3. Primary analysis based on all cases 32 (39)as randomised

Comparability of treatment and control groups at entry1. Large potential for confounding or 28 (34)

not discussed

2. Confounding small; mentioned but not 12 (14)adjusted for

3. Unconfounded; good comparability of 43 (52)groups or confounding adjusted for

For hip or other appendicular skeleton fracture0. Not applicable 34 (41)

3. X-ray confirmation of diagnosis 13 (16)

4. No confirmation of diagnosis 36 (43)

For vertebral fracture0. Not applicable 3 (4)

1. Inadequately described method 27 (33)

2. Radiological method: uses anterior/ 12 (14)posterior height ratio

3. Radiological method: uses anterior, 41 (49)middle and posterior height in criteria OR reports radiologically confirmed clinical events only


40

interventions, and capture of primary endpoints,including the definition of incident vertebralfracture. Where possible, sensitivity analyses were used to explore the implications of such heterogeneity.

Vertebral fractureThe data reviewed above suggested thatbisphosphonates, calcitonin, fluoride, raloxifene,and possibly vitamin K2, reduced the risk ofvertebral fracture in patients with osteoporosis or osteopaenia. Calcium also appeared to beeffective in patients with low calcium intakes.There was no evidence that vitamin D derivatives,

oestrogen, oestrogen-like molecules, anabolicsteroids, protein supplements or brisk walkingreduced this risk.

The question arises whether efficacy differs inpatients recruited into a study with or withoutprevalent vertebral fractures at trial entry. This isan important issue since differences would haveimplications for the assumptions used for healtheconomics modelling in established osteoporosis.Information was available for bisphosphonate,fluoride and SERMs. For none of the treatments(Table 26) was there a significant difference whenpatients were stratified according to the presence

TABLE 25 Summary of quality assessment: mean scores

Mean scores Year of publication

1980–89 1990–94 1995–2000 All trials

All publicationsFully blinded randomisation 1.13 1.25 1.36 1.29

Blinded assessment of outcome 1.93 1.92 1.95 1.94

Withdrawals 2.07 2.13 2.30 2.20

Comparability of groups at baseline 1.87 2.04 2.36 2.18

Confirmation of hip or other appendicular skeleton fracture 1.66 1.55 1.50 1.53

Appropriateness of method of diagnosis of vertebral fracture 2.00 1.96 2.37 2.18

Total a 60% 61% 66% 64%

Excluding abstractsFully blinded randomisation 1.13 1.27 1.44 1.29

Blinded assessment of outcome 1.93 2.00 2.06 2.01

Withdrawals 2.07 2.23 2.42 2.29

Comparability of groups at baseline 1.87 2.09 2.61 2.30

Confirmation of hip or other appendicular skeleton fracture 1.66 1.60 1.57 1.59

Appropriateness of method of diagnosis of vertebral fracture 2.00 2.05 2.59 2.30

Total a 60% 63% 71% 67%

a Expressed as a percentage of total possible score

TABLE 26 RRs (with 95% CIs) of vertebral fracture compared with placebo or no treatment in patients with and without priorvertebral fracture

Intervention RR (95% CI)

All patients Prior fracture No prior fracture

Vertebral fractureBisphosphonates 0.569 (0.493 to 0.656) 0.575 (0.490 to 0.675) 0.558 (0.387 to 0.805)

Fluoride 0.794 (0.647 to 0.973) 0.686 (0.544 to 0.864) 0.250 (0.078 to 0.797)

SERMs 0.661 (0.579 to 0.755) 0.674 (0.581 to 0.780) 0.575 (0.436 to 0.757)

Non-vertebral fractureBisphosphonates 0.824 (0.745 to 0.913) 0.813 (0.693 to 0.954) 0.889 (0.761 to 1.039)

Fluoride 0.975 (0.782 to 1.217) 0.998 (0.788 to 1.268) 0.412 (0.130 to 1.308)

SERMs 0.920 (0.792 to 1.068) No data No data


41

or absence of prevalent vertebral fractures. A trendwas observed in the case of fluoride but this didnot reach statistical significance. It would thereforeappear appropriate to use information based on all patients eligible for study.

A further heterogeneity of potential importance is the criterion used to define incident vertebralfractures. It is well recognised that the lessstringent criteria increase the apparent incidenceof vertebral fracture – but at the expense of a high false-positive rate.112 For example, if inbiological reality, ten fractures occurred in theplacebo arm and five in the treatment arm, thiswould give a RR reduction of 50%. If the same trial was contaminated with false-positives (forexample, three in each arm), the apparent efficacy would fall to 38%. In this review, thecriteria used to define vertebral fracture variedbetween studies. In some, a minimum reduction of 15% in vertebral height was required, in some20%, and in others the definition used was notspecified. It has been demonstrated that the use of a 20–25% definition rather than a lower figuresuch as 15% will increase the power of a study byreducing the number of false-positives.186 In thisreview, studies that used a 20% definition pro-duced results more favourable to the interventionthan those which used a 15% definition. Theavailable data are summarised in Table 27. Nocomparative data were available for oestrogens (all 15% criteria) or calcium (all 20% criteria).

Non-vertebral fractureBisphosphonates were the only intervention that wasdemonstrated by the meta-analyses reported aboveto reduce the risk of non-vertebral fracture generallyin patients with osteoporosis or osteopaenia. How-ever, no intervention has been demonstrated toprotect against non-vertebral fracture in subjectswithout prior fracture (see Table 26).

When only the hip fracture data were taken intoaccount, bisphosphonates were again the only

intervention that was demonstrated by the meta-analyses reported above to reduce the risk of suchfracture generally in patients with osteoporosis orosteopaenia. However, if these results are sub-divided according to an individual’s fracture status,bisphosphonates were not shown to have a signifi-cant effect in reducing the risk of hip fracture even in those with prior fracture (see Table 10).This is probably due to the decrease in power of the analysis, since there was little difference in the control group estimate between thecombined and disaggregated analysis.

Duration of studyThere is growing awareness that the efficacy ofinterventions with time may be non-linear. If so,this has important implications for modelling when the modelling period exceeds the durationavailable from RCTs. The available information,shown in Table 28, suggested that transients couldoccur – in the sense that mid-point estimates ofefficacy appear to be less favourable the longer the duration of study, for both vertebral and non-vertebral fractures. In most instances, it was notpossible to consider year-by-year effects. Were thisto show a waning of treatment effect with time, it would have important implications for healtheconomics modelling, in that caution would beneeded when modelling treatment duration forlonger than the duration of the RCTs.

Comparison of treatmentsThe midpoint estimates for efficacy are sum-marised in Table 29. For the purposes of vertebralfracture frequency, diagnostic criteria were usedthat included a 20% decrease in vertebral height to define an incident fracture where available. For forearm fracture, estimates were used whenavailable (e.g. bisphosphonates). In the absence of data, and for humeral fracture, it was assumedthat efficacy would be equivalent to our estimatesof efficacy for non-vertebral fractures. When thecentral estimate of efficacy was markedly affectedby quality, high-quality studies were used in


TABLE 27 Efficacy of interventions according to the criteria used to define incident vertebral fractures

Intervention RR (95% CI)

15% decrease in vertebral height 20% decrease in vertebral height

Bisphosphonates 0.628 (0.506 to 0.779) 0.526 (0.435 to 0.637)

Vitamin D derivatives 1.266 (0.717 to 2.235) 0.459 (0.149 to 1.414)

Calcitonin 0.670 (0.477 to 1.004)a 0.308 (0.133 to 0.838)

Fluoride 1.107 (0.819 to 1.496) 0.350 (0.253 to 0.486)

Raloxifene 1.166 (0.832 to 1.635) 0.596 (0.516 to 0.688)

a Criteria not specified


42

sensitivity analyses (for forearm fractures, in thecase of bisphosphonates). Estimates of forearmand humeral fractures are included since they areincorporated into the health economics model.

It was evident that bisphosphonates, calcitonin,calcium, fluoride and SERMs all significantlydecreased vertebral fracture frequency. Althoughthe risk estimate for vitamin D derivatives wasquantitatively similar, the CIs cross unity. Withrespect to hip fracture, significant effects were

seen only with the bisphosphonates, although thecentral point estimates for vitamin D derivativesand for calcitonin were similar but not significant.Neither fluoride nor SERMs appeared to conferprotection against hip fracture. Bisphosphonateswere the only agent that significantly decreasedappendicular fractures.

Onset and offset of actionIn this review, it was assumed that the endpointestimate of efficacy derived from RCTs would apply

TABLE 28 RR of vertebral and non-vertebral fracture in intervention compared with control arm, according to duration of study (or,when data available, by year within study)

RR (95% CI)

Length of study 1 year 2 years Over 2 years

Vertebral fractureBisphosphonates No data 0.481 (0.260 to 0.888) 0.571 (0.495 to 0.659)

Vitamin D derivatives 0.347 (0.086 to 1.403) 1.113 (0.680 to 1.824) No data

Calcitonin 0.077 (0.001 to 4.274)a 0.308 (0.113 to 0.838) 0.685 (0.456 to 1.029)

Oestrogen-like molecules 0.200 (0.003 to 13.783) 0.515 (0.190 to 1.397) No data

Fluoride 0.540 (0.273 to 1.070) No data 0.653 (0.524 to 0.816)

SERMs 1.166 (0.832 to 1.635) No data 0.596 (0.516 to0.688)

Non-vertebral fractureBisphosphonates 0.518 (0.300 to 0.894) 0.915 (0.688 to 1.216) 0.827 (0.739 to 0.924)

Vitamin D derivatives 1.104 (0.004 to 277.68) 1.371 (0.338 to 5.559) No data

Calcitonin 0.556 (0.069 to 4.491)a 0.525 (0.204 to 1.350) 1.062 (0.043 to 26.059

Oestrogen-like molecules No data No data No data

Fluoride No data 1.234 (0.809 to 1.883) 0.892 (0.686 to 1.161)

SERMs No data No data 0.920 (0.792 to 1.068)

a Data derived from 6-month study in subjects with steroid-induced osteoporosis

TABLE 29 Comparative effects of different agents to reduce fracture risk

Intervention Site of fracture

Vertebraa Hip Forearm Humerusb

Alendronate 0.544 0.611 0.866c 0.825

Bisphosphonates 0.526 0.672 0.833 0.824

Calciumd 0.55 – – –

Vitamin D derivativese 0.459c 0.249c 1.042c 0.193c

Calcitonin 0.308 0.681c 0.947c 0.530c

Fluoride 0.350 1.778c 0.776c 0.975c

Oestrogen 0.583cdf – – –

SERMsd 0.596 1.141c 0.887c 0.920c

a Assessed as 20% decrease in vertebral height (see Table 27)b Assumed to be equivalent to non-vertebral fracture (see Table 25)c Confidence estimate crosses unityd Single studye For alfacalcidol alonef Criterion for vertebral fracture = 15%


43

to the 5-year treatment interval to be used forhealth economics modelling. As mentioned earlier,a longer time frame would provide a less securebasis since, for some treatments, transients inefficacy are found. Our own analysis suggested that vertebral fracture efficacy is more marked in the early years of treatment compared with later effects in the relevant RCTs. A notableexception may occur in the case of oestrogens and hip fracture risk. In a population-based,case–control study of oestrogen use, each year of use was associated with a risk of decrease of 6% (95% CI, 3 to 9).197

A key assumption relating to the long-termeffectiveness of an intervention is the duration forwhich an effect persists after stopping treatment –a concept that has been termed offset time.44

A great deal of uncertainty surrounds the offset of therapeutic effect once treatment has stopped.Relatively rapid offset of effects has been observedwith calcium, calcitonins and vitamin D meta-bolites. In the case of calcium supplements andvitamin D derivatives, prospective studies of theoffset time have shown that bone mass at 2–3 yearsafter treatment is the same as that in untreatedindividuals.123,198,199 Such data suggest that nofurther gains can be expected from treatments that had been stopped 3 years earlier.

In contrast, it is a widely held view that no catch-up loss of bone occurs when treatment with oestrogens is stopped. The results of onestudy suggested a catch-up loss over a period of 2–3 years200 but the results of another prospectiverandomised study suggested otherwise.201 In thelatter study, patients were followed for a 12-yearperiod. One group received no treatment, anotherreceived HRT continuously and a third groupreceived HRT for 5 years, followed by 7 years of no treatment. In the latter group, the value ofBMD lay between that of the other two groups, at a value consistent with no accelerated bone loss or an offset time equal to infinity.

Nevertheless, a number of epidemiological studieshave suggested that a slow catch-up of bone lossoccurs over 15–20 years after stopping treatmentwith oestrogens or oestrogen-like agents, so that by the age of 80 years, the effect of a treatment for 5 or 10 years at the age of 50 years has all but disappeared. The evidence is derived fromepidemiological studies of BMD as well as hipfracture rates.55,197,202–205 In a recent populationcase–control study, use of oestrogens within the last 12 months was associated with a 62% decreasein hip fracture risk (RR = 0.38; 95% CI, 0.26

to 0.56). In individuals who had taken HRT 13–60 months previously, the risk reduction was48% (RR = 0.52; 95% CI, 0.26 to 1.04). Use of HRT more than 5 years previously was associatedwith a 25% risk reduction (RR = 0.75; 95% CI, 0.52 to 1.07).197

In the case of the bisphosphonates, the offset of effect has not been fully characterised. Thecessation of treatment is associated with anincrease in skeletal markers of resorption andformation but bone loss does not appear to occur immediately. In one study, average lossesover 3 years, after stopping treatment withalendronate for 3 years, were comparable to those in placebo-treated patients206 but the timecourse of change was not reported. A sustainedeffect of bisphosphonates was observed following a short course of alendronate in the treatment of osteoporosis.207,208 Recent studies of the use of pamidronate and alendronate have suggested,however, that bone loss may eventually resume at an accelerated rate.209,210 Similar findings havebeen reported for risedronate.211 The offset times for anabolic regimens have not beencharacterised but bone loss occurred shortly after stopping treatment with fluoride.212 Bonemass appeared to be preserved in oestrogen-treated women after stopping treatment withparathyroid hormone but continuing their HRT (R Lindsay, Helen Hayes Hospital, New York;personal communication, 1998). In contrast, offset times appeared to be shorter after stoppingtreatment with calcium and with alfacalcidol.44,198

An important impact of offset on therapeutic effect had been previously noted on fracturesprevented.43,214,215 A recent study showed theprofound impact of different assumptionsconcerning offset time on cost-effectiveness.44

Several health economics analyses have examinedthe effects of intervention once treatment isstopped; most related to the effects of oestro-gens.216 Weinstein assumed that oestrogensdecreased fracture risk to 0.33 and, after stoppingtreatment, the RR increased to 0.5 for a durationthat equalled the exposure time of the activetreatment.7 Similar assumptions have been madeby other investigators.2,217,218 Others assumed a slowoffset of effect so that a 10-year treatment at themenopause had a slow offset of effect up to the age of 75 years or more.219,220 The most optimisticscenario assumed an infinite offset time.10

In this study, an offset time of 5 years hasconservatively been assumed, except for calcium



44

and calcitonin for which an offset time of 3 yearswas assumed. An increased offset time of 10 yearswas examined in a sensitivity analysis.

Side-effectsAs noted above, the various interventions reviewedhere varied in their associated effects, both adverseand, in some cases, beneficial. The adverse effectsmost commonly found in association with thedifferent interventions are set out in Table 30. It is important to recognise that a systematic review of side-effects has not been undertaken and only studies eligible for the meta-analysis have been reviewed.

In addition, other studies have shown thatoestrogen increases the risk of breast cancer,venous thromboembolic events, gall-bladderdisease and, unless opposed by progestogen,endometrial cancer. In addition to gastrointestinalcomplaints, ipriflavone may be associated with skinreactions and, to a lesser extent, neurological,musculoskeletal and cardiovascular symptoms.

Some of the interventions studied also have associ-ated effects that are beneficial. Thus, calcitonin,ipriflavone and nandrolone have been associatedwith reductions in the intensity of osteoporoticpain and with improved mobility. In one study,72

fluoride also appeared to be associated with asignificant reduction in back pain. Oestrogen canreduce menopausal symptoms such as hot flushesand may provide protection against colorectalcancer. Raloxifene may have a preventive role inrelation to breast cancer. Brisk walking reducesboth high-density lipoproteins and obesity.

ContinuanceAs noted above, for each intervention there wasconsiderable variation between studies in terms of the percentage of patients treated with theactive intervention and who completed theprotocol. This information is summarised for each intervention in Table 31.

Compliance, in terms of the number of individualswho continued to take the medication and theproportion of medication which they had taken,was reported in few studies specifically. Hence, the information provided is too heterogeneous to summarise further.

TABLE 30 Adverse effects most commonly reported in the reviewed studies

Intervention Adverse effect

Bisphosphonates Upper gastrointestinal events

Vitamin D derivatives Hypercalciuria; hypercalcaemia

Calcitonin (injected) Hot flushes; gastrointestinal complaints

Calcitonin (intranasal) Hot flushes; gastrointestinal complaints; rhinitis and minor local nasal or respiratory disorders

Calcium Gastrointestinal symptoms

Oestrogen Pelvic congestion; unspecified oestrogen-related adverse events

Oestrogen-like molecules Gastrointestinal complaints

Anabolic steroids Hoarseness; voice lowering; increased facial hair

Fluoride Lower extremity pain, bone lesions and incomplete fractures; osteomalacia;gastrointestinal complaints

SERMs Venous thromboembolism; hot flushes

Protein supplements None

Vitamin K2 No information

Exercise Falls

TABLE 31 Continuance (percentage of patients completingprotocol) as reported by reviewed studies

Intervention Continuance (%)

Bisphosphonates 58–95

Vitamin D derivatives 65–91

Calcitonin (injected) 50–100

Calcitonin (intranasal) 57–84

Calcium 71

Oestrogen 83–92

Oestrogen-like molecules 56

Anabolic steroids 91

Fluoride 50–78

SERMs 78–over 90

Protein supplements 73

Vitamin K2 No information

Exercise 19


45

It was to be expected that continuance andcompliance with medication would be higher in the context of an RCT than in real life. Forexample, although in the studies reviewed hereover 80% of women taking oestrogen completedthe various trials, other studies have reportedconsiderably lower compliance in patientsprescribed HRT for osteoporosis, with reports of 36% and 49% compliance after 1 year and 61% after 6 months to 1 year.154–156

Discussion

Evidence of efficacy has been identified in thisreview for only some of the interventions studied:bisphosphonates for vertebral and non-vertebralfracture; calcitonin, calcium, fluoride and raloxi-fene for vertebral fracture only. However, failure to demonstrate the efficacy of the remaininginterventions and of calcitonin, calcium, fluorideand raloxifene in relation to non-vertebralfracture, may reflect the small size and shortduration, as well as the inappropriate reporting of fracture outcomes, of the studies that weresuitable for meta-analysis.

In addition, it was decided to study only RCTs inwhich information was provided on the skeletalstatus of individuals in terms of BMD or priorfragility fracture. Thus studies directed to thegeneral population or individuals characterised athigh risk by other means have been ignored. Thishas resulted in some important omissions. Finally,epidemiological information that provided a lowerlevel of evidence has been ignored. The extent towhich the totality of these omissions tempers viewsconcerning efficacy is reviewed briefly below.

BisphosphonatesThe bisphosphonates, as a class, have been shownby RCTs to reduce vertebral and non-vertebralfractures, including hip fracture. An exception for the individual bisphosphonates is the apparentlack of efficacy of etidronate on hip fracture risk.The studies reviewed were undertaken in womenwith vertebral fracture and were not powered toassess hip fractures. However, an RCT publishedafter our cut-off date indicated that risedronatedecreased the risk of hip fracture by 30%.221 Inwomen with osteoporosis, the effect was moremarked (RR = 0.6; 95% CI, 0.4 to 0.9). Also,epidemiological investigation has shown that theuse of etidronate is associated with a significantreduction in hip fracture risk.54 It is notable thatthe efficacy estimate is greater for vertebralfracture than for appendicular fractures. Thus, if

the association is causal, there appears to be littledifference in efficacy between bisphosphonates.

Vitamin D and calciumCalcium is widely available throughout the worldand is the major non-HRT intervention used inosteoporosis.55 It is commonly used in combinationwith vitamin D (vitamin D2 or vitamin D3). Noinformation is provided in this review relating to these agents, with the exception of two studieswith calcium, one of which reported a significantdecrease in vertebral fracture frequency inestablished osteoporosis. One epidemiologicalstudy has found that the risk of hip fractureappears to be decreased in women takingpharmacological amounts of calcium;55 the effects persisted even after adjusting for potentialconfounding factors. Calcium supplements weretaken on average at age 70 years, whereas theaverage age of hip fracture was 75 years.

A controlled prospective study in the elderly has shown that the combined use of calcium andvitamin D significantly decreased the frequency of hip fracture.35,222 Over an 18-month follow-upperiod, 204 non-vertebral fractures occurred in the calcium group compared with 355 fractures in the placebo group. The decrease in bothfemoral and other non-vertebral fractures wassignificant. It should be noted that patients weredrawn from sheltered accommodation and thepossibility exists that some had coexisting vitaminD deficiency. Nevertheless, the dose of vitamin Dused was physiological and the findings lendcredibility to the retrospective studies suggestingthat calcium with or without physiological doses of vitamin D decreases the risk of hip fractures.BMD was not systematically assessed in this study;hence, it was not included in our primary analysis.However, low BMD was found in a sample ofpatients surveyed and, for this reason, an effect on non-vertebral fractures was included in asensitivity analysis. The risk of hip fracture was significantly reduced (RR = 0.738; 95% CI,0.600 to 0.908), as was the risk of all non-vertebral fractures (RR = 0.793; 95% CI, 0.687 to 0.917). The latter estimate has been used for forearm and humeral fractures in a sensitivity analysis.

A further study with vitamin D alone was partlybased in the community and partly drawn fromnursing home occupants.223 In this study, anintramuscular injection of 150,000–300,000 unitswas given annually to individuals aged 75 years or more. Treatment resulted in a significantdecrease in fracture frequency.



46

These findings contrasted with those reported for the use of a modest dose of vitamin D alone(400 IU daily) in 2600 elderly men and womenfrom Holland.224 After a follow-up period of 42 months, 58 hip fractures had occurred in thevitamin D-treated group and 48 in placebo-treatedpatients. Other peripheral fractures occurred withequal frequency. It is possible that the apparentlack of efficacy relates to the low dose of vitamin Dused and the high nutritional status with respect to calcium in Holland. Studies of calcium andvitamin D alone and in combination are presentlybeing conducted in the UK, and many helpdistinguish the effects of these agents.

Vitamin D derivativesThe present review suggests that the 1-alphahydroxylated derivatives of vitamin D do notsignificantly decrease the risk of fractures. Mid-point estimates were decreased for vertebral andhip fracture but not significantly so (see Table 29).Epidemiological studies would also suggest that any effect on hip fracture risk is not significant.55

Tilyard and colleagues have published the resultsof a large prospective study.124 This prospective,randomised but open-label study was undertakenin New Zealand in more than 600 postmenopausalwomen with vertebral fracture (see page 37). The study was not included in our primary analysisbecause the trial was a comparison of calciumalone versus calcitriol alone. Vertebral fracture rate did not increase in the first year of treatment;thereafter, there was a significant difference infracture rate between patients receiving calcitrioland those receiving calcium (RR estimate = 0.31).

CalcitoninThe present study indicates that calcitonin signifi-cantly decreases vertebral fracture risk. Evidencethat other fractures are decreased is not provided.This relates to the small numbers of patientsenrolled to examine this end point. Some supportfor an effect on hip fracture risk is provided by acase–control study undertaken in SouthernEurope.55 The RR of hip fracture associated withthe use of calcitonin was 0.63 (95% CI, 0.44 to0.90). The study also had positive and negativecontrols. Oestrogens were shown to be associatedwith a decrease in risk whereas fluoride wasassociated with a modest increase in risk.

OestrogensAlthough oestrogens are widely considered to be the treatment of choice for the prevention ofosteoporosis, the RCT evidence that this is associ-ated with a decrease in fracture risk is wanting. In case–control studies, HRT has been consistently

associated with a decrease in appendicularfractures including hip fracture.55,151,197

Efficacy was supported by a small prospectiverandomised study,225 in which HRT decreased the risk of non-vertebral fractures by 71% (RR =0.29; 95% CI, 0.10 to 0.90). Intention-to-treat(ITT) analysis gave an RR of 0.44 (95% CI, 0.21 to 0.93). The study was not included in our meta-analysis because women were not selected on thebasis of low BMD. Evidence of efficacy was alsosupported by prospective population cohortstudies.151,226,227 In this study, estimates for efficacyon hip and forearm fractures were taken fromCauley and colleagues,227 since the population was older (65 years or more), the study wasprospective and women with a history ofosteoporosis were assessed separately. The RR for hip fracture was 0.86 (95% CI, 0.42 to 1.75)and for forearm fracture 0.32 (95% CI, 0.13 to0.78) in current users of HRT with a history ofosteoporosis. For humeral fractures, the RR for all non-spinal fractures was used (RR = 0.63;95% CI, 0.45 to 0.89). The estimate for hipfracture efficacy is conservative. For example, results from a case–control study suggested anefficacy of 50% or more.197

New data have recently become available from a study undertaken in the USA (the Women’sHealth Initiative). This showed a decrease infracture risk but an increased risk of adversecardiovascular events.

In addition to osteoporosis, HRT has many multi-system effects that could be of considerableconsequence in the health economics setting.

HRT is widely used to control menopausalsymptoms. Most women experience symptoms but only a minority find them a problem. For example, in a UK survey, 57% of womenexperienced hot flushes but only 22% consideredthem to be a problem.228 HRT is very effective in controlling menopausal symptoms, and morethan 40 RCTs relating to its use have beenundertaken, most of which show significantbenefit.229 Although the treatment of menopausalsymptoms has been shown to be cost-effective,2

these effects were not incorporated in this report, since the proportion of patients withestablished osteoporosis and menopausalsymptoms is very small.

The greatest potential benefit of HRT is oncardiovascular disease. Observational studiesshowed a consistent decrease in risk. A recent


47

meta-analysis found a summary RR of 0.70 (95% CI, 0.67 to 0.75). For opposed oestrogen, the RR was 0.66 (95% CI, 0.53 to 0.84).230 Sincehealthy women preferentially take HRT, theseobservational studies are likely to be biased in itsfavour194 and the extent of this bias is uncertain. In a review of 22 short-term RCTs in which cardio-vascular events were recorded, the risk of cardio-vascular disease was higher (RR = 1.39) in womentaking HRT than in a control group.232 Morerecently, the effect of HRT in women at risk fromcoronary artery disease was examined in a largeRCT.153 Overall, no effect was found (RR = 0.99;95% CI, 0.80 to 1.22), although some protectionwas afforded in the later phase of the study.

In our base case, no protective effect of HRT oncardiovascular disease has been assumed but theresults of the meta-analysis by Barrett-Connor andGrady were used in the sensitivity analysis.230 Inconcordance with the observational studies, riskprotection was assumed to disappear as soon as treatment stopped.233,234

For HRT, the adverse effects of greatest concernare related to breast cancer risk. The clinicalinformation available was largely derived fromepidemiological studies, which suggested that HRT is associated with a small increase in risk.Moreover, the association is plausible. For thisstudy, a recent meta-analysis of 51 epidemiologicalstudies was used, comprising 52,700 women withand 108,400 women without breast cancer.149 Therisk appeared to increase by 2.3% for each yearthat HRT was used. This is equivalent to the excessrisk associated with a late menopause. Thus, a 5-year exposure might be assumed to increase the RR by 11.5% (i.e. RR = 1.12). In this study and conservatively, the collaborative group’sestimate of long-term use (RR = 1.35; 95% CI, 1.21 to 1.49) has been used, based on the use of HRT for at least 5 years. It was assumed that the effect disappeared as soon as treatment was stopped.235

HRT has been associated with a large number of other beneficial and adverse effects that werenot included in this analysis. Beneficial effectsincluded reductions in colonic cancer (RR = 0.76;95% CI, 0.70 to 0.82), rectal cancer (RR = 0.81;95% CI, 0.72 to 0.92)146 and Alzheimer’s disease(OR = 0.71; 95% CI, 0.53 to 0.96), and increasedsurvival.148,236 Adverse effects included an increasedrisk of endometrial cancer with unopposed use,152

which was substantially reduced by cyclic progesto-gen.145 An increased risk of venous thrombo-embolic disease has also been reported.237–239

The RR increases were 3.5 (95% CI, 1.8 to 7.0)238

and 3.6 (95% CI, 1.6 to 1.8),239 respectively, but theabsolute risk increase was small (16.5/100,000 inwomen aged 45–64 years). Other possible adverseeffects include gall bladder disease, vaginalbleeding, bloating and breast tenderness.151

Oestrogen-like moleculesThere was no supplementary information onfracture rates with these compounds.

Anabolic steroidsThere have been no prospective randomisedstudies, in which fracture outcomes were theprimary end-point, to determine whether anabolicsteroids reduce fracture frequency. In a retro-spective case–control study, the use of anabolicsteroids in women was associated with a marked(RR = 0.6) but not significant decrease in the RRof hip fracture.55 Analysis of these data for Italy –where anabolic steroids are widely used – showed a significant effect.240

Fluoride and thiazidesOur own analysis indicated that fluoride canreduce vertebral fracture frequency but has nosignificant effect on hip fracture risk. The lack ofeffect on non-vertebral fractures was consistentwith epidemiological studies.55 Our conclusionsconcerning vertebral fracture effects were atvariance with those of a recent meta-analysis,241

in which fluoride had no apparent effect onvertebral fracture risk (RR = 0.87; 95% CI, 0.51 to 1.46). The disparity arises from differences inthe studies excluded or included. In this review,studies in normal individuals, non-randomisedstudies and those without fracture end-points were excluded. Also, in the meta-analysis,241

several studies were not included for reasons that were not apparent;73,79,80 studies in men were also excluded.173

In this review, no studies were identified thatshowed antifracture efficacy for thiazides. Inseveral longitudinal studies, rates of bone loss were reported to have decreased in men takingthiazides for hypertension.242 In case–controlstudies, the use of thiazides was associated with a significantly decreased risk of hip fracture.243–245

A meta-analysis of these case–control studiessuggests that current use of thiazides is associatedwith a decrease in fracture risk of borderlinesignificance (RR = 0.82; 95% CI, 0.62 to 1.08).246

SERMsThe data reviewed here on osteoporotic fracturerepresent the totality of information available for



48

raloxifene. The results of a large prospective study undertaken in women with breast cancersuggested that tamoxifen decreased the risk of hipfracture.204 This observation may be importantsince the studies in raloxifene were not powered to demonstrate effects on appendicular fractures.

Raloxifene has been shown to have some HRT-like effects on biochemical markers of cardio-vascular disease (see page 35). It decreased totalcholesterol by 3–6% and low-density lipoproteincholesterol by 4–10%. High-density lipoproteincholesterol and triglyceride concentrations did not change significantly. A meta-analysis of total

cholesterol lowering in the setting of primary and secondary cardiovascular disease suggestedthat cardiac events decreased by 2% for every 1%reduction in serum cholesterol.210 On this basis, in the sensitivity analysis it has been assumed that raloxifene might decrease cardiovascular risk by 20%.

As in the case of HRT, other beneficial or adverseeffects have not been considered, including anincrease in incidence of venous thrombo-embolicevents. The increase in risk (RR = 3.1; 95% CI, 1.5to 6.2) is comparable to that derived fromobservational studies with HRT.


49

In addition to information on efficacy, side-effects and compliance associated with

interventions, economic evaluation requires infor-mation on additional components that include:

• the incidence of osteoporotic fractures• the prevalence of established osteoporosis• the risk of fracture associated with osteoporosis• the risk of fracture associated with established

osteoporosis• the mortality associated with established

osteoporosis• the interactions between osteoporosis and other

health states• the direct and indirect costs of osteoporotic

fractures• costs of treatment and monitoring• utilities for osteoporotic fractures with which

to quality-adjust years of life saved.

Whenever possible the data used to populate thehealth economics model have been derived from a UK information base. The sources of data, theirlimitations and the assumptions that derive fromthem are reviewed here. The uncertainties that areinherent in the estimates provide a rationale forsubsequent sensitivity analysis.

Osteoporotic fracture

DefinitionThe definition of an osteoporotic fracture is not straightforward. An approach that is widelyadopted is to consider low-energy fractures asbeing caused by osteoporosis. This has the merit of recognising the multifactorial causation offracture. However, with high-energy trauma, frac-tures are more likely in osteoporotic individualsthan those without osteoporosis.248 There is also a disparity between low-energy fractures andfractures associated with reductions in BMD.249

The classification is therefore incomplete.

A further approach is to characterise a fracture as osteoporotic only in the presence of osteo-porosis, as defined by the T-score and WHOcriteria,13 or to identify the types of fracture that increase in frequency as the BMD falls. The association of several different fracture types

with BMD was investigated in a large NorthAmerican survey – the Study of OsteoporoticFractures (SOF)249 – and is the approach that someresearchers have used to exclude some fracturetypes as not being due to osteoporosis. An addi-tional criterion is to examine the pattern of frac-tures with age. A rising incidence of fractures with age does not provide evidence that thefracture type is caused by osteoporosis, since arising incidence of falls could also be a cause. In contrast, a lack of increase in incidence with age is reasonable presumptive evidence that afracture type is unlikely to be osteoporosis-related.An indirect arbiter of an osteoporotic fracture isthe finding of a strong association between thefracture and the risk of classical osteoporoticfractures at other sites. Vertebral fractures, forexample, are a very strong risk factor for sub-sequent hip and vertebral fracture.84,250,251

Irrespective of the methods used, opinions differabout the inclusion or exclusion of different sitesof fracture. Those that are included here comprisehip fractures, vertebral fracture coming to clinicalattention; fractures of the distal forearm and proxi-mal humerus. These comprise the most commonsites of osteoporotic fracture but, nevertheless,exclude fractures at other sites, such as pelvis,distal femur, rib and tibia. The proportion offractures accounted for by other osteoporoticfractures in Sweden is summarised in Table 32.41

The proportion of all fractures accounted for bythe sites chosen represent the majority (65–74%),depending on age. They also represent the majorcause of morbidity. The fraction, however, dependscritically on age. Thus, between the ages of 50 and55 years, fractures of the hip, forearm and spineaccount for 57% of the morbidity from all osteo-porotic fractures but account for 84% over the age of 85 years. In health economics modelling,the exclusion of other fractures will underestimatethe benefits from treatment, although it should be acknowledged that the evidence for efficacy of any treatment at these sites is wanting.

A further assumption is that all fractures at aparticular site that is included are caused byosteoporosis. This is clearly an oversimplification.Since some important fracture sites (e.g. pelvis)


Chapter 4

Epidemiology, costs and utilities


50

have been excluded, this may be offset by theassumption that all fractures at an included site are due to osteoporosis. An alternative approach is to quantify, by expert opinion, the proportion of fractures at each site caused by osteoporosis. This approach is used in Switzerland252 and theUSA253,254 to characterise the burden of disease; however, it too is arbitrary and based on as many assumptions.

Fracture risksThe fracture risks used in this report are derived as far as possible from UK sources. There havebeen several recent surveys reporting fracture rates in the UK.17,25,255,256 (van Staa T, Procter and

Gamble Pharmaceuticals, Staines, UK: personalcommunication, 2001; Cooper C, MedicalResearch Council, Southampton, UK: personalcommunication, 2000) The data from van Staa arefrom general practice research data. The annualincidence of osteoporotic fractures from recentsurveys is given in Table 33. For hip, forearm andproximal humeral fracture rates, the data fromSinger and colleagues,25 based on a population inEdinburgh, were used. These data were preferredto those from Cardiff, reported by Johanssen andcolleagues,256 since more fractures were analysed(15,293 versus 6467). Hip fracture rates in theseries from Singer and colleagues25 were midwaybetween the estimates of Johansen and colleagues

TABLE 32 Proportion of osteoporotic fractures (%) at the sites shown and the proportioned utility loss by age in women fromSweden41 (Reproduced with permission from Osteoporosis International)

Fracture type Age range (years)

50–54 55–59 60–64 65–69 70–74 75–79 80–84 85+

Proportion of fracturesVertebra 15 13 19 16 20 17 13 11Hip 4 7 11 15 21 26 37 36Forearm 39 37 36 26 23 16 13 10Shoulder 12 10 8 13 10 10 6 8Other 30 33 26 30 26 31 31 35

Proportion of disutilityVertebra 20 14 17 14 14 10 6 5Hip 21 33 42 52 61 68 78 77Forearm 11 9 7 5 4 3 2 1Shoulder 5 3 2 3 2 2 1 1Other 43 41 32 26 19 17 13 16

TABLE 33 Annual fracture risk in women by age (%)

Age range Annual fracture risk (%)(years)

Hip Wrist Shoulder

Singer, Johansen, Van Singer, Johansen, Van Singer, Johansen, Van 1998a 1997a Staab 1998a 1997a Staab 1998a 1997a Staab

50–54 0.041 0.05 0.02 0.255 0.343 0.24 0.058 0.035 0.06

55–59 0.05 0.125 0.05 0.374 0.531 0.40 0.085 0.077 0.11

60–64 0.083 0.125 0.08 0.467 0.531 0.48 0.136 0.077 0.12

65–69 0.157 0.36 0.15 0.573 0.667 0.57 0.126 0.09 0.17

70–74 0.485 0.36 0.28 0.699 0.667 0.62 0.246 0.09 0.22

75–79 0.707 1.34 0.55 0.697 0.917 0.71 0.306 0.299 0.28

80–84 1.437 1.34 1.03 0.749 0.917 0.80 0.372 0.299 0.35

85–89 2.761 4.14 1.71 1.001 1.37 0.84 0.362 0.628 0.45

90+ 3.851 4.14 2.33c 0.919 1.37 0.89c 0.391 0.628 0.47c

a See list of trials meeting inclusion criteria (page 125)b Van Staa, Procter and Gamble Pharmaceuticals, Staines, UK: personal communication, 2001c 90–94 years


51

and van Staa, but were broadly comparable.Similarly, for forearm and shoulder fractures,fracture rates in Singer and colleagues lay midwaybetween those of Johansen and colleagues and van Staa. Overall, the differences in risk were lessthan the ranges found within other countries. Forexample, estimates of hip fracture risk betweenseries vary two-fold within Norway and Turkey.26,257

There are few data relating to vertebral fractureincidence in the UK, and none are ideal. Theproblem is due in part to incomplete informationand in part to the criteria used to define a verte-bral fracture. Vertebral fractures that come toclinical attention are less common than fracturesdiagnosed from change in vertebral shape obtain-ed by morphometric techniques. With respect tosymptomatic fractures, data were available from the Trent region but comprised only those patientswho required hospital admission,258 thus under-estimating the incidence considerably. Vertebralfracture rates were also available from generalpractice research data but their validity was un-certain. Age-stratified estimates of prevalence andincidence of vertebral deformities were availablefrom the Chingford study but the sample size andage ranges were small.259 (van Staa, Procter andGamble Pharmaceuticals, Staines, UK: personalcommunication, 2001) More substantial data fromseveral UK centres are available, although not yetpublished, from the EPOS (European ProspectiveOsteoporosis Study) database. These too aremorphometrically-diagnosed deformities andprovide incidence rates that will be substantiallygreater than the incidence of clinically diagnosedfractures (Table 34). In this study, it was consideredpreferable to estimate the incidence of clinicallydiagnosed fracture, since it is these patients whoare most likely to be identified for treatment.Moreover, the QALY estimate ascribed to vertebralfractures pertains to clinically diagnosed fractures.

For these reasons, vertebral fracture rates wereimputed from data available from Malmö,

Sweden, on the incidences of hip and vertebralfractures that come to clinical attention.24 It was assumed that the ratio of the incidence ofvertebral fracture and hip fractures in Malmöwould be comparable to the ratio of vertebralfracture incidence in the UK (unknown) and hipfracture incidence in the UK (Edinburgh). Therates are shown in Table 35. Though data arelimited, there appears to be a consistency in thepattern of different osteoporotic fractures in theWestern world.41 Thus, the assumptions that were made concerning proportionality appearreasonable. The rates shown (see Table 35) aresubstantially lower than those derived from radio-graphic surveys but higher than those reportedfrom the General Practice Framework. Forexample, in women aged 70–74 years, the annualrate used was 0.68% compared with 0.12% fromthe General Practice Framework and 1.2% usingmorphometric criteria. The rate used thus liesbetween the two UK direct estimates.

It is important to note that the QALY estimate forvertebral fracture is conservative (see page 65),and that morphometrically diagnosed fracturesgive rise to current morbidity and are associatedwith a high risk of future fractures. For this reason,cost-effectiveness was examined also, using themorphometrically-derived incidence rates in asensitivity analysis and assuming from our calcu-lations that 24% of morphometric deformitiescome to clinical attention.

There are several other uncertainties relating tothe risks that were used. Regional estimates maynot be representative of the UK and the use ofthese estimates in modelling future events gave rise to even greater concerns, for several reasons.

1. It was assumed that over 10 years (the time frameused in the base case), the risk of fracture willnot change in the population. The secular trendin hip fracture risk appears to have flattened inthe UK29 but if age-specific rates decrease in thefuture, the impact of treatments on fractureburden will be overestimated. Secular trends forother fractures are not documented in the UKbut age- and gender-specific incidence appears to be increasing in some countries.260

2. It was also assumed that the mortality hazard doesnot change over 10 years. However, mortality has continued to decrease and this is likely tocontinue in the future.22 Failure to take accountof these trends will underestimate the impact oftreatments on the numbers of fractures saved.

3. The fracture rates used are drawn frompopulation samples over a limited period


TABLE 34 Incidence of vertebral fracture (rate/1000/year) inEurope and the UK using morphometric criteria

Age range Europe UK(years)

Men Women Men Women

50–54 1.75 4.2 0 5.4

55–59 6.7 5.8 2.9 6.4

60–64 6.5 11.0 3.0 11.8

65–69 8.7 14.5 6.9 5.1

70–74 9.4 18.5 4.2 11.7

75–79 14.8 33.0 0 34.6


52

(1 year). The fracture rates given will includethose individuals with a first fracture and aminority who have previously sustained afracture at that site. Thus, the risk of firstfracture is overestimated. The overestimate is greater in the elderly than in the young.24

Thus there are factors that variously overestimateand underestimate fracture risk. In this study, theseare not considered further since the cancelling outof the sources of error is likely to provide a morereasonable estimate than corrections based onuntestable assumptions.

BMD and fracture risk

Gradients of riskA number of prospective studies examined the risk of fracture as a function of BMD. In general,the lower the BMD value, the greater the risk offracture. The increase in fracture risk is approxi-mately doubled for each SD decrease in BMD.Thus, for an individual with a BMD value one SD lower than the average BMD for a given age, the fracture risk is about twice that for anindividual with the average BMD for that age. The gradient of risk, however, varies according to the site of assessment and the technique used.The most extensive meta-analysis was undertakenby Marshall and colleagues.16 For absorptiometrictechniques, the gradient of risk depended on the site of measurement as well as the technique. For example, BMD measurements by dual-energy X-ray absorptiometry to predict hip fracture arebetter when the measurements are made at the hip rather than at the spine or forearm (Table 36).Thus, an individual with a T-score of –3 SD at the hip would have a 2.63 or greater than 15-fold higher risk than an individual with

a T-score of zero SD. In contrast, the same T-score at the spine would yield a much lower risk estimate – approximately a four-fold increase(1.63). Similarly, spine measurements predict spinefractures more accurately than measurementsmade at other sites.

The gradient of risk is highest for hip fractureprediction from measurements at the hip. Also,measurements at the hip predict all fractures aswell as measurements at other sites. For thisreason, the proximal femur is the preferred andrecommended site for diagnostic use,15 and it isassumed here that this is the site that would beused for diagnostic purposes. For humeral frac-tures, no data are available and the gradient of risk with which all fractures were predicted(1.6/SD) was assumed. There is little difference in predictive power between measurements madeat the femoral neck or in the region of total hip. In this study, the femoral neck was chosen since a UK reference range was available.

The gradients of risk that have been assumed are very similar to those found in the SOF study –the largest prospective study of fracture risk inwomen.9 In this study, the risk of hip fractureincreased 2.6-fold for each SD decrease in hipBMD – a value that is identical to the midpointestimate of the meta-analysis by Marshall andcolleagues16 (see Table 36). At this site, the gradi-ent of risk for forearm fractures was 1.6, and forvertebral fracture 1.9, compared with 1.7 and 1.7 in Marshall and colleagues’ meta-analysis.

Fracture risk in osteoporosisThe computation of fracture risk from BMD inindividuals demands a knowledge of the change ofBMD in the ageing population, so that the averageT-score can be computed according to age.

TABLE 35 Fracture risk (%) by age at the sites shown

Age range (years) Fracture risk (%)

Hip Vertebral Wrist Proximal humerus

50–54 0.041 0.108 0.255 0.058

55–59 0.050 0.144 0.374 0.085

60–65 0.083 0.127 0.467 0.136

65–69 0.157 0.111 0.573 0.126

70–74 0.485 0.683 0.699 0.246

75–79 0.707 0.600 0.697 0.306

80–84 1.437 0.755 0.749 0.372

85–89 2.761 1.223 1.001 0.362

90+ 3.851 1.155 0.919 0.391


53

In the approach used here, the relationship be-tween Z-score and T-score at different ages wasexamined, using conversion factors calculated froma published survey of normal women in London.261

This study measured BMD at the femoral neck. Thesame approach was used by the US National Osteo-porosis Foundation (NOF) but, in their report,9 theyoung normal reference range was not defined andthe data were derived from measurements at thetotal hip site rather than the femoral neck.

The relationship between T-score and Z-score is given in Table 37. For example, for the average90-year-old individual, Z-score = 0. This would beequivalent to a T-score of –2.72 SD. Similarly, a 70-year-old individual with an average BMD (Z-score = 0) would have a T-score of –1.65 SD, and someone with a Z-score of –1.0 would have a T-score of –2.65 SD (that is, –1 minus –1.65).

From these data, the annual risks of fracture can be computed for individuals with a T-score of –2.5 SD, that is, at the diagnostic threshold for osteoporosis (Table 38).

Consider, for example, a 70-year-old woman withan average BMD for her age (Z-score = 0 SD).

The T-score is –1.65 SD, which in turn is 0.85 SDabove the threshold for osteoporosis. The annualrisk of hip fracture for a 70-year-old woman withosteoporosis, therefore:

= average risk for a 70-year-old individual × (2.6)0.85

= average risk × 2.253= 1.093%

The risk of a vertebral fracture:= average risk × (1.8)0.85

= average risk × 1.648= 1.1256%

The risk of a forearm or humeral fracture:= average risk × (1.6)0.85

= average risk × 1.491= 1.042% and 0.367%, respectively.

The annual risk of fracture for women at thethreshold for osteoporosis is presented by age in Table 38. Note that for a risk at a given age (e.g. 70 years), the T-score conversion is used for that age and multiplied by the risk of the age interval (e.g. 70–75 years). Thus, the risk isoverestimated at the start but, as the patient agesin the model, the risk is underestimated.

The approach used is an oversimplification. It is assumed that the increase in risk due to osteo-porosis is a multiple of the RR and the populationrisk. The oversimplification arises because BMD is normally distributed in a population at a givenage, whereas there is an exponential relationshipbetween BMD and fracture risk. Thus, individualswith an average BMD have a lower-than-averagerisk of fracture.262 Conversely, the average fracturerisk is found in individuals with a lower-than-average BMD. For this reason, the calculations of fracture risk at the threshold for osteoporosisare overestimated. An indication of the over-estimate in the case of hip fracture is presented in Table 39.

The risk of fracture in women at the threshold for osteoporosis differs from that of a populationof osteoporotic patients, since few will have a


TABLE 36 RRs (with 95% CIs) of fracture in women for a 1 SD decrease in BMD (absorptiometry) below the age-adjusted mean16

(Reproduced with permission from the BMJ Publishing Group)

Site of measurement RR (95% CI) of fracture

Forearm Hip Vertebral All

Distal radius 1.7 (1.4–2.0) 1.8 (1.4–2.2) 1.7 (1.4–2.1) 1.4 (1.3–1.6)Femoral neck 1.4 (1.4–1.6) 2.6 (2.0–3.5) 1.8 (1.1–2.7) 1.6 (1.4–1.8)Lumbar spine 1.5 (1.3–1.8) 1.6 (1.2–2.2) 2.3 (1.9–2.8) 1.5 (1.4–1.7)

TABLE 37 Converting Z-scores to T-scores

Age NOF9 a Frost et al., 2001261

(years)T-score = T-score =

Z-score minus: Z-score minus:

50 0.37 0.60

55 0.69 0.86

60 1.01 1.12

65 1.29 1.39

70 1.56 1.65

75 1.84 1.91

80 2.11 2.18

85 2.31 2.44

90 2.52 2.70

a Conversions at ages 55, 65, 75 and 85 years have beenestimated using linear interpolation


54

T-score at the threshold value. From the distri-bution of BMD and the change with age, the risksof fracture can be compared.24 The differences in risk are shown in Table 40. Thus, the RR isunderestimated in the average osteoporoticpatient, and this underestimate increases with age – from 65% at age 50 years to 158% at 84 years. For this reason, sensitivity analyses havebeen undertaken by increasing the stringency of the T-score: for example, a T-score of –3.5 rather than –2.5 SD.

Fracture risk in establishedosteoporosisRisk ratiosA large number of studies have examined the risk of fracture following a fragility fracture.Fracture risk is increased over and above thatexplicable on the basis of age or BMD. Theinterrelationship between prior and subsequentfractures has recently been assessed by meta-analysis using a random effects model to derivesummary estimates of RR.263 The risk estimatesderived from peri- or post-menopausal women are shown in Table 41 adjusted for age but notBMD. There have been no studies that examinedthe risk of a forearm fracture following a hipfracture and an RR of 1.4 was used, equivalent to the lowest RR between fractures. The relevantrisk functions are summarised in Table 42.

The analysis computes that (for example) anindividual with a prior spine fracture has a risk of a further spine fracture that is 4.4-times greater

TABLE 38 The estimated annual risk of fracture (%) at the sites shown in the female population and in women with a T-score of –2.5 SD

Age range Fracture risk (%)(years)

Hip Vertebra Wrist Shoulder

Popu- Osteo- Popu- Osteo- Popu- Osteo- Popu- Osteo-lation porosis lation porosis lation porosis lation porosis

50–54 0.041 0.252 0.108 0.330 0.255 0.623 0.058 0.142

55–59 0.050 0.240 0.144 0.378 0.374 0.809 0.085 0.184

60–64 0.083 0.310 0.127 0.286 0.467 0.893 0.136 0.260

65–69 0.157 0.453 0.111 0.213 0.573 0.966 0.126 0.212

70–74 0.485 1.093 0.683 1.126 0.699 1.042 0.246 0.337

75–79 0.707 1.242 0.600 0.849 0.697 0.914 0.306 0.401

80–84 1.437 1.950 0.755 0.911 0.749 0.870 0.372 0.432

85–89 2.761 2.896 1.223 1.260 1.001 1.025 0.362 0.371

90+ 3.851 3.181 1.155 1.026 0.919 0.836 0.391 0.356

TABLE 40 RR of hip fracture in women at, and at and below, the threshold (T) for osteoporosis (reproduced with permission fromElsevier Science)

Age (years) RR of hip fracture B/A A/B

(A) when T = –2.5 SD (B) when T ≤ 2.5 SD

50 2.9 4.8 1.65 0.6155 2.4 4.1 1.71 0.5860 1.9 3.4 1.79 0.5665 1.5 2.9 1.93 0.5270 1.2 2.5 2.08 0.4875 0.95 2.1 2.21 0.4580 0.74 1.8 2.43 0.4184 0.62 1.6 2.58 0.39

TABLE 39 Risk of hip fracture in Swedish women at thethreshold for osteoporosis262 (reproduced with permission fromElsevier Science)

Age (years) RR compared with:

Average BMD Population

50 4.6 2.9

60 3.0 1.9

70 1.9 1.2

80 1.2 0.74


55

than that of an individual of the same age butwithout fracture. It is assumed thereafter that anindividual with a T-score of –2.5 SD with a priorvertebral fracture has a 4.4-fold greater risk thanan individual without such a fracture but the same T-score.

This assumption is an oversimplification for severalreasons. Firstly, the risk estimates are not adjustedfor BMD, which is likely to overestimate the risk by up to 20%. For example, in the SOF study of7238 women aged 65 years or more, those with aprevalent vertebral deformity had a 4.1-fold higherrisk of a further vertebral fracture. When adjustedfor BMD, the RR was 3.8. Similarly, in women withvertebral fracture, the age-adjusted risk of hip frac-ture was 1.9, which fell to 1.8 after adjustment.85

Secondly, some downward adjustment of RR is required since, apart from BMD, the risks arerelative to those without prior fracture rather than to the general population. The adjusted RR approximates to:

RR/(p.RR + (1 – p))

where p is the prevalence of the risk factor (priorfracture) and RR the unadjusted value.264 Forexample, the RR of hip fracture in the presence

of a prior vertebral fracture is given as 4.4. If, at a given age, the prevalence of a prior vertebralfracture is 5% (0.05), the population RR is:

4.4/(0.05 × 4.4 + (1–0.05)) = 3.76

When the prevalence is 10% of the population, theRR falls to 3.28. These overestimates of fracture riskare offset because the risk of subsequent fracturevaries with time after fracture. In the case of verte-bral fracture requiring hospitalisation, risks offurther osteoporotic fractures are markedly in-creased immediately after a fracture and tail off to those observed in the meta-analysis after 12–18months.265 Also, the increase in RR varies accordingto age and is markedly higher in the young than inthe elderly (Table 43).265 These short-term RRs morethan offset the overestimates arising from othersimplifications. Thus, the assumptions used wereconservative, particularly in younger individuals.

The simplifications used are amenable to sensitivityanalysis and have been accommodated here.

Fracture riskThe data permit the calculation of fracture risk in established osteoporosis conditional upon thesite of prior fracture. For example, the risk of hipfracture in a woman aged 70 years is approximately


TABLE 41 Risk of fracture at the sites shown according to the site of a prior fracture263 (reproduced with permission from theAmerican Society of Bone and Mineral Research)

Site of prior fracture RR (95% CI) of subsequent fracture

Distal forearm Spine Proximal Hip Pooledhumerusa

Forearm 3.0 (2.0 to 5.3) 1.7 (1.4 to 2.1) 2.4 (1.7 to 3.4) 1.9 (1.6 to 2.2) 2.0 (1.7 to 2.4)Spine 1.4 (1.2 to 1.7) 4.4 (3.6 to 5.4) 1.8 (1.7 to 1.9) 2.3 (2.0 to 2.8) 1.9 (1.7 to 2.3)Humerusa 1.8 (1.3 to 2.4) 1.9 (1.3 to 2.8) 1.9 (1.3 to 2.7) 2.0 (1.7 to 2.3) 1.9 (1.7 to 2.2)Hip 1.4 (–b) 2.5 (1.8 to 3.5) 1.9 (–c) 2.3 (1.5 to 3.7) 2.4 (1.9 to 3.2)Pooled 1.9 (1.3 to 2.8) 2.0 (1.6 to 2.4) 1.9 (1.6 to 2.2) 2.0 (1.9 to 2.2) 2.0 (1.8 to 2.1)

a Assumed to be equivalent to a ‘minor fracture’ from the meta-analysisb No studiesc One study

TABLE 42 The increased risk of subsequent fracture following a prior fracture

Prior fracture site Increased risk of subsequent fracture

Hip Vertebra Wrist Shoulder

Hip 2.3 2.5 1.4 1.9

Vertebral 2.3 4.4 1.4 1.8

Wrist 1.9 1.7 3.3 2.4

Shoulder 2.0 1.9 1.8 1.9


56

0.5%. With a T-score of –2.5 SD, the RR increasesby 2.60.85 = 2.25, equivalent to a hip fracture riskof 1.13%. In a woman with the same T-score but a prior spine fracture, the risk of hip fracture isincreased 2.3-fold (see Table 42), giving a hipfracture risk of 2.26%. Further examples are given in Table 44.

In the absence of specific information, it wasassumed that bone loss occurs at the same rate inuntreated patients as in the general population.

Consequences of fracture

Death due to a hip fractureExcess mortality is well described after hip fracture.In the first year following hip fracture, the mor-tality risk in women varies from 2.0 to > 10 depend-ing on age (Table 45).22 It classically follows a bi-phasic pattern, with a sharp increase 6 months to 1 year after the event and thereafter decreasing,but remaining higher than in the general popu-

lation.266 Since hip fracture patients have highcoexisting morbidity, poor pre-fracture health islikely to contribute to the excess mortality. Case–control studies adjusting for pre-fracture morbidityindicate that a substantial component can beattributed to comorbidity.267,268 Irrespective of the attribution, it is not possible to determine thequantum of excess mortality that would be avoidedin the absence of hip fracture. It can be arguedthat the acute increment in mortality over the first6 months is reversible by avoiding fracture. Duringthis period, the excess mortality risk was estimatedto be 3.35 (95% CI, 1.50 to 7.47) compared with asubsequent risk of 1.30 (95% CI, 0.85 to 1.98).267

The data that were used for death following hipfracture were unpublished data from the secondAnglian audit of hip fracture, in which deaths upto 90 days were recorded (Table 46).269

Parker and Anand266 estimated that 33% of deaths1 year after hip fracture were totally unrelated tothe hip fracture, 42% possibly related and 25%

TABLE 43 Incidence (rate/1000) of hip fracture admissions in the general female Swedish population and of patients at 6 months orat 4 years following hospitalisation for vertebral fracture265 (Reproduced with permission from Osteoporosis International)

Age range (years) General Incidence at RR Incidence at RRpopulation 6 months (95% CI) 4 years (95% CI)

50–54 0.36 6.0 (4.2–8.4) 16.7 3.8 (2.7–5.4) 10.6

55–59 0.83 10.7 (8.6–13.3) 12.9 6.8 (5.4–8.6) 8.2

60–64 1.46 19.0 (16.9–21.5) 13.0 12.2 (10.5–14.0) 8.4

65–69 2.53 29.9 (27.0–33.3) 11.8 19.1 (16.8–21.8 7.5

70–74 5.05 39.0 (35.9–42.2) 7.7 24.9 (22.3–27.8) 4.9

75–79 10.73 50.7 (47.7–53.9) 4.7 32.3 (29.4–35.5) 3.0

80–84 18.52 65.9 (62.4–69.5) 3.6 42.1 (38.4–46.1) 2.3

85–89 32.2 85.7 (80.5–91.2) 2.7 54.7 (49.7–60.2) 1.7

TABLE 44 Risk of fracture (%) at the sites shown in established osteoporosis with a T-score of –2.5 SD at the ages shown

Site of prior fracture Risk of subsequent fracture at the sites shown

Forearm Spine Shoulder Hip

Age 50–54 yearsForearm 2.06 1.06 1.49 1.18

Spine 0.46 1.45 0.59 0.76

Shoulder 0.26 0.27 0.27 0.28

Hip 0.35 0.63 0.48 0.58

Age 70–74 yearsForearm 3.44 1.77 2.50 1.98

Spine 1.58 4.95 2.03 2.59

Shoulder 0.62 0.64 0.64 0.67

Hip 1.53 2.73 2.08 2.51


57

directly related. These figures were not, however,stratified by age or gender and have been assumedto be constant for all ages.

The raw data for deaths at 90 days were multipliedby 67% to subtract the number of deaths that were assumed to be unrelated to the hip fracture;possibly related deaths were attributed to the hipfracture, which may overestimate the death rate.

As deaths beyond 90 days following hip fracturewere not recorded in the Anglian audit, an assump-tion was made that no further deaths caused by hipfracture occurred after 90 days. It is noted that thismay underestimate the mortality rate due to hipfracture, as it appears from a graph in Parker andAnand’s paper266 that, in relation to deaths at 90days, there would be approximately 40% of addi-tional deaths in the period 91–365 days. However,some redress has been made by assuming that the42% of deaths possibly related to hip fracture weredirectly due to the fracture. Overall, the assump-tions attribute 48% of all deaths within the first year as related to hip fracture.

The 0% death rate in nursing homes at ages 60–69 years shown in Table 46 is empirical butappears to be lower than expected. Indeed, therelative mortality hazard is higher in younger agegroups than in the elderly. In the first year afterfracture, the RR in women aged 50–74 years was3.2 compared with 1.6 at the age of 85 or moreyears.270 Even higher risks were found in Sweden(see Table 45). This is unlikely to have a significantimpact as, in the model, an assumption was madethat patients can only enter a nursing homefollowing a prior hip fracture, and the absolute risk of fracture at the age of 60 years is low.

First entry to nursing home following a hip fractureThe model has been populated with unpublisheddata from the second Anglian audit of hip fracture

(Institute of Public Health, 1999) summarisedelsewhere.271 The percentage of hip fractures that resulted in a first admission to a nursing home were 4% for those aged 60–69 years and70–79 years, 12% for those aged 80–89 years and 17% for those aged 90 years and over.

Death due to vertebral fractureSeveral studies have shown an increase in mortalityfollowing vertebral fracture.272–274 In one study,women with one or more vertebral fractures had a 1.23 fold greater age-adjusted mortality rate(95% CI, 1.10 to 1.37). Unlike hip fracture, thereis no acute excess.272,274 It is notable that low BMDis also associated with excess mortality275,276 but the degree of increased mortality after vertebralfracture is greater than that expected from lowBMD. In the model, an excess associated with low BMD has been assumed (see later) but noadditional mortality from vertebral fracture.

These studies used morphometric definitions of vertebral fracture. In contrast, other studies that examined mortality after vertebral fractureusing clinical criteria have shown more marked


TABLE 45 Mortality by age in Sweden (1994) in the general female population and in the year following hip fracture22 (Reproducedwith permission from Osteoporosis International)

Age (years) Population (rate/1000) Hip fracture (rate/1000) Excess mortality (RR)

50 2.25 35.8 15.955 3.07 25.7 8.460 5.06 54.8 10.865 8.23 39.6 4.870 15.53 97.1 6.375 25.8 80.8 3.180 47.1 199.5 4.285 83.4 166.1 2.0

TABLE 46 Estimated 1-year mortality rates due to hip fracture

Place of residence Age range Mortality rate(years) at 1 year (%)

Community 60–69 4a

Community 70–79 6

Community 80–89 11

Community 90+ 16

Nursing home 60–69 0



Nursing home 90+ 23

a 6% in Anglian audit but adjusted downwards from Swedishnational data (see Table 45)


58

increases in mortality.273,277 (Johnell O, Departmentof Orthopaedics, Malmo, Sweden: personalcommunication, 2001; McCloskey EV, WHOCollaborating Centre for Metabolic Bone Disease,Sheffield, UK: personal communication, 2001) In one study in Australia, vertebral fractures inwomen were associated with an age-standardisedrisk of 1.92 (95% CI, 1.70 to 2.14)273 and, inanother study, the risk was more than eight-foldhigher.277 In the present model, incidence rates of clinically diagnosed fractures were used, somortality may be underestimated. It was alsoassumed, perhaps conservatively, that inter-vention had no effect on mortality.

Death due other fracturesIt was assumed that there was no increase inmortality from forearm or humeral fracturesconsistent with published surveys.273,274,277

Breast cancer and cardiovasculardiseaseContracted breast cancerThe incidence of breast cancer was taken from cancer registrations278 and is summarised in Table 47, assuming a population as reported by the Office of National Statistics.279

Two large cohort studies showed that osteoporosisor low BMD was associated with a lower incidenceof breast cancer.280,281 Conversely, approximately70% of breast cancer cases occurred in womenwith a BMD in the two highest quartiles (Table 48).The study by Cauley and colleagues280 indicatedthat the risk of breast cancer was increased by 1.34 per one SD increase in BMD (measured at the proximal radius). This equates approximatelyto a 0.75 RR per one SD below the populationaverage BMD. For this reason, the breast cancer

risk was adjusted downwards (see Table 47). Thus,for example, the RR of breast cancer for womenaged 70–75 years has been calculated as:

0.268 × 0.750.94

It was assumed that interventions that increasedthe BMD of a patient would not change the risk ofbreast cancer adjusted because of a low T-score.

Death due to breast cancerAs the model is an individual patient-based simu-lation, the data required are the probabilities ofdeath due to breast cancer in each year followingdiagnosis. As such, standard summary data such as total death rates due to breast cancer per year are inappropriate.

The data that were used are the 5-year survivalrates for the years 1986–90 in England andWales.279 The 5-year survival rate was 68%.Comparison of 1-, 5- and 10-year survival ratesshowed a steep decline in mortality, followed by a flattening of the death rate after 5 years. It wasassumed that any patient who survived beyond 5 years would not die from breast cancer.

For the 32% who died within the 5-year period, itwas assumed that the survival period was 2 years.The model will record the patient as dying in theyear that breast cancer was diagnosed but theQALYs accrued are doubled and the annual costsincurred by the patient are doubled for that year.

Death due to CHDThe number of deaths from CHD (InternationalClassification of Disease codes 410–414) was takenfrom mortality statistics from the Office of NationalStatistics.278 The population figures were from thesame source. The estimated annual risks arepresented in Table 49.

Non-fatal CHD eventsThe data for CHD events were derived from thetotal of CHD deaths and ratios of deaths to eventsas presented by Volmink and colleagues.282 Theratios of non-fatal definite myocardial infarction(MI) and non-fatal possible MI to fatal definite

TABLE 47 The annual incidence of breast cancer by age

Age group Annual incidence Average(years) of breast cancer population

Osteoporotic population50–54 0.245 0.133

55–59 0.277 0.164

60–64 0.319 0.208

65–69 0.257 0.181

70–74 0.269 0.205

75–79 0.284 0.235

80–84 0.320 0.286

85+ 0.362 0.343

TABLE 48 The relationship between breast cancer and BMD

Source Quantile of BMD

1 2 3 4

Cauley, et al., 1996280 9 20 37 34

Zhang, et al., 1997281 13 19 20 48


59

MI plus fatal possible MI and fatal unclassifiedcoronary death were 2.26 in the 50–64 years’ agegroup and 0.82 in the 65–79 years’ group. Formodelling purposes, it was assumed that theseratios were for 55–59 years and 70–74 years,respectively. Estimates of the ratios for other age bands were made assuming a linear changebetween 55–59 years and 70–74 years.

These data do not contradict the ratio of 2.6 ofnon-fatal MIs to fatal MIs reported by Stampferand colleagues,234 who analysed the data forwomen between the ages of 30 and 63 years.

It was assumed that the MI ratios were applicableto CHD ratios.

The estimated risk of a non-fatal CHD incident isgiven in Table 49.

Death from other causesThese data were taken from interim life tables(Table 50)283 and adjusted for deaths due to CHDand breast cancer in the general population.

Several studies have shown an increased mortalityassociated with low BMD of similar magnitude de-rived from measurements at the radius or heel.275,276

At the radius, the increase in RR was 1.22 per 1 SDdecrease in BMD adjusted for age,275 and this valuewas used in the model. These data are shown inTable 50. When a patient died from other causes, the costs and QALYs from the previous year werehalved for the year in which the patient died.

It was assumed that those interventions thatincreased the BMD of a patient would not changethe risk of death due to other causes adjusted due to a low T-score.

Health state utility valuesIt was necessary to identify the best available utilityestimates for health states associated with theconsequences of established osteoporosis and itstreatment for use in the model. The health statesused in the model included healthy osteoporosis,established osteoporosis, hip fracture, vertebralfracture, wrist fracture, proximal humerus fracture,breast cancer and CHD. Previous economic evalu-ations of the prevention and treatment of osteo-porosis relied on the use of assumptions or judge-ments obtained from expert panels such as therecent review undertaken by the NOF,9 rather than using empirical evidence to value these health states. This is recognised as one of the main weaknesses of work in this area.216,284

Recently there have been a number of studieseliciting health state valuations for many of thesestates using recognised preference-based measuresof health-related quality of life (such as theEuroQol-5 dimension (EQ-5D) or the HealthUtility Index (HUI)-III) or direct preferenceelicitation techniques, such as time trade-off(TTO) or standard gamble (SG). It was thereforedecided not to rely on assumptions for this modelbut to make use of empirically derived figureswhen available.

Identifying the key studiesThis review drew on papers identified from a series of systematic searches. These includedsearches of papers reporting economic evaluationof the prevention and treatment of osteoporosis,and those reporting on quality of life associatedwith the main fracture states. Information was alsosought on breast cancer and CHD, since severaltreatments for osteoporosis may affect these out-


TABLE 49 The annual incidence of non-fatal events and deathsdue to CHD

Age range Annual incidence of CHD (%)(years)

Non-fatal events Deaths

50–54 0.072 0.026

55–59 0.144 0.064

60–65 0.240 0.135

65–69 0.364 0.280

70–74 0.442 0.541

75–79 0.317 0.941

80–84 0.000 1.637

85–89 0.000 2.449

90+ 0.000 3.251

TABLE 50 The annual death rate due to other causes

Age range Risk of Risk of death adjusted(years) death for low BMD

(%) (%)

50–54 0.243 0.371

55–59 0.402 0.576

60–65 0.680 0.915

65–69 1.206 1.536

70–74 2.015 2.429

75–79 3.271 3.733

80–84 5.497 5.940

85–89 9.250 9.62

90–94 14.896 14.896

95–99 22.796 22.796

100+ 27.008 27.008


60

comes. Studies were identified through searches of electronic databases, hand-searching, citationsearching, reference list checking and contactingof individuals known to researchers involved in the study. Details of the methods are presented in appendix 6.

A total of 1132 papers were found. Their abstracts were examined in order to identify those likely to be relevant to this study, whichreduced the number to 173. A further sifting was undertaken to identify those papers pre-senting health state values likely to be relevant to this model. Most of the 173 papers wereconcerned with measuring quality of life in general and did not present preference-basedhealth state values. Six published papers werefound on fracture states and seven on breastcancer and CHD. The former group of papers is complete, while those for CHD and breastcancer represent only a selection. These 13 papers contained 36 health state values, since a number of papers reported on more than one state, values by different groups ofrespondents and/or used more than one valu-ation technique. These published papers weresupplemented by two unpublished studies,285,286

of which the first has now been published.

ResultsThe health state values were found to differconsiderably from the assumptions used inpreviously published economic evaluations in this area (Table 51). The value for vertebralfractures used by the NOF of 0.97, for example,compares with values ranging from 0.31 to 0.80.These empirical estimates used a recognisedpreference elicitation procedure but there was a considerable range of values for each healthstate. This range reflects a number of differencesin the derivation of the estimates including: the source of values, what is being valued, thevaluation technique and the anchor states used in the valuation task. The selection of estimates for the model involved both technical and value judgements that are discussed below.

Methodological issues in selectinghealth state valuesThe source of valuesThe studies differed in terms of the samples beingasked to undertake the valuations. Some estimateswere obtained by asking patients to value their ownhealth states, others by asking samples of patientsto value hypothetical descriptions of the states. Inother studies, values were elicited from samples ofprofessionals (e.g. Hutton and colleagues, 1996)287

or representative samples of the generalpopulation.288 Having patients value their ownhealth states has the advantage of avoiding theneed to describe health states and may ensure that they have a better understanding of theimpact of the state on their lives. However, it has the disadvantage that it limits the source of values to current patients. It has been arguedthat, for the purposes of informing resource allo-cation, the values of society at large are requiredand, hence, studies that used a representativesample of the general population would be more appropriate.289

What is being valued?Respondents were asked to value speciallyconstructed vignettes to describe each healthstate290 or to use generic preference-basedmeasures such as the EQ-5D.292 The genericpreference-based measures come with a set of values already obtained from a generalpopulation sample. The state-specific approach has the potential advantage of being more rele-vant and sensitive to the condition than genericmeasures.292 The disadvantage is that the descrip-tions may only represent a proportion of thehealth states found in a sample of patients, and it is not clear how representative they would be of patients with the condition.

Generic preference-based measures have theadvantage of being administered to a sample of patients and thus represent the variation inhealth states found in the population. To use these estimates, however, it is necessary to establish that the study sample accurately reflects the population used in the model, in terms of variables such as age, severity and other background variables that are likely to affect a patient’s state of health. This can bedifficult to achieve in practice, since most data in this area have been collected from patients recruited into trials with strict criteria for inclusion, which may not result in the mix of patients used in the model (and typically seen in the NHS) and in different countries.

Adjusting for the age composition of the samplesThe health state values did not cover the full range of age groups used in the model. Somestudies were limited to one age group (e.g. Brazier and colleagues, 2000);286 others were based on such small numbers that it was notpossible to estimate reliable age-specific values.One approach to extrapolating the findings


61


TABLE 51 Empirical estimates of utility values for osteoporosis-related health states

Source Utility values Methods

Health state description Valuation technique Source of values

‘Healthy’ osteoporoticNoF review9 1.0 Judgement Panel of experts

Kind, et al., Age range (years) EQ-5D for general TTO valuations of EQ-5D General population 1998296 45–49 0.840 population; hence, hypothetical states with (n = 3381)

50–54 0.850 includes all sources of full health and dead as55–59 0.802 morbidity found in reference states60–64 0.829 this population65–69 0.80670–74 0.74775–79 0.73180–85 0.69985+ 0.676

‘Established osteoporosis’, i.e. history of broken wrist, spinal or hip fractureGabriel, 0.84 (± 0.29) Patients who experienced TTO valuation of own health Patients (n = 75;et al., 1999288 non-traumatic vertebral anchored by best imaginable mean age, 76 years)

fracture in last 5 years but for age and deadnot multiple fractures

0.43 (± 0.40) Valuation of hypothetical TTO valuation of hypo- Non-fracture cases state constructed from thetical health state anchored attending clinic clinician views and focus by current health and dead in last 2 years groups – these include transformed using valuation (n = 199; mean reference to future risk of own health against perfect age, 68 years)

health and dead

Hip fractureNOF review9 First year: 0.3817 Assumes quality-of-life Judgement Expert panel

reduction from acute care,rehabilitation, home care,GP visits and ambulance

Subsequent Assumes distribution years: 0.855 across disability states

Gabriel, 0.68 (0.18) 37 patients who had hip frac- HUI-II valued by SG HUI-II parents of schoolet al., 1999288 ture in last 5 years completed (estimated from children from Hamilton,

HUI-II (mean age, 76 years) transformation of VAS) Canada (n = 203)

0.61 (0.08) Above patients completed QWB valued by VAS Representative sampleQWB scale of general population

of San Diego, USA

0.72 (0.16) Above patients valued their VAS Patients (n = 37)own health states

0.70 (0.41) Above patients valued their TTO anchored by perfect Patients (n = 37)own health states health and dead, where

perfect health is bestimaginable for their age

0.65 (0.45) Patients’ own valuation of TTO anchored by perfect Patients (n = 33;their hip fracture states, health and dead, where mean age, 76 years)which they regarded as perfect health is bestworse than hypothetical imaginable for their age‘disabling’ fracture state

0.28 (0.37) Hypothetical ‘disabling’ TTO anchored by own health Recent clinic hip fracture state state and dead, and latter attendees who

transformed using their have never had a valuation of own health state fracture (n = 198,(itself anchored against best mean age, 68 years)imaginable for health and dead)

continued


62

TABLE 51 contd Empirical estimates of utility values for osteoporosis-related health states



Hip fracture contdSalkeld, 0.31 Based on description of life TTO anchored by hypothetical Older people at risk of et al., 2000290 (interquartile after ‘good’ hip fracture health state in someone of fracture (n = 194; mean

range, 0.0–0.65) similar age to respondent age, 81 years)and death

Brazier, at 6 months: 39 patients completed EQ-5D TTO valuations of EQ-5D General population et al., 2000286 0.49 (0.32) before fracture and then at hypothetical states with full (n = 3381)

at 12 months: 6 and 12 months after fracture health and dead as reference 0.48 (0.38) (mean age, 76 years) states

Confined to nursing home due to hip fractureNOF review9 0.4 Nursing home Judgement Expert panel

Salkeld, et al., 0.05 Based on description of life TTO anchored by hypothetical Older people at risk of 2000290 (no range given) after ‘bad’ hip fracture that health state of someone of fracture (n = 194)

included being in nursing home similar age to respondent and death

Vertebral fractureNOF review9 0.97 Assumes 33% experience no Judgement Expert panel

change; 57 % quality of life reduced by 0.5 for 1 month;10% experience complete loss,then 0.5 loss for 7 weeks

Gabriel, 0.80 (0.16) 94 patients who had vertebral HUI-II valued by SG HUI-II parents of school et al., 1999288 fracture in last 5 years (estimated from trans- children from Hamilton,

completed HUI-II formation of VAS) Canada (n = 203)

0.66 (0.09) Above completed QWB QWB valued by VAS Representative sample of general population of San Diego, USA

0.76 (0.17) Above patients valued VAS Patients (n = 94)their own state

0.81 (0.32) Above patients valued their TTO anchored by perfect Patients (n = 94)own state health and death, where

perfect health is best imaginable

0.68 (0.4) Patients’ own valuation of their TTO anchored by perfect Patients (n = 24)fracture state, which they health and death, where perfect regarded as worse than health is best imaginablemultiple vertebral fracture state

0.31 (0.38) Hypothetical multiple TTO anchored by own health Recent clinic attendees vertebral fracture state state and death, and latter who had never had

transformed using their fracture (n = 199)valuation of own health state (itself anchored against best imaginable for health and death)

Oleksik, 0.744 (0.231) 130 patients who had EQ-5D for general population; TTO valuations of et al., 2000285 experienced radiographically hence, includes all sources of EQ-5D hypothetical

confirmed fracture in last morbidity found in this states with full health 5 years completed EQ-5D population and dead as reference

states

continued


63


TABLE 51 contd Empirical estimates of utility values for osteoporosis-related health states



Wrist fractureNOF review9 1st year, 0.96; Assumes 0.7 for 7 weeks Judgement Expert panel

subsequent years,0.98 Assumes long-term

dependency for 2% of patients with QoL reduction to 0.7

Dolan, 0.982 EQ-5D completed by 50 wrist TTO valuations of EQ-5D General population et al., 1999297 fracture attendees (mean age, hypothetical states with (n = 3381)

72 years, range, 52–91) at full health and dead as outpatient clinic at first and reference statesfinal visit (average 48-day inter-val). Implied QALY loss overyear, assuming linear progressionbetween initial and last assess-ment, 0.018 (0.014)

Breast cancerHutton, et al., 0.62 Stable disease SG using McMaster UK oncology nurses 1996287 0.33 Progressive disease pingpong method (n = 30). Similar figures

0.84 Partial response to therapy obtained from nurses in Hypothetical health state three other countriesdescriptions constructed from multi-disciplinary group; no variance data given

Grann, 0.89 (interquartile No detail offered on TTO – no protocol Convenience sample et al., 1998a range, 0.86–1.00) descriptions used detail provided (n = 54)

De Haes, 0.65 Health state described as Crude VAS values, subject to n = 27, healthcare et al., 1991b 3 months to 1 year after TTO power function workers and cancer

mastectomy experts0.17 Terminal illness

CHDOldridge, 0.717–0.767 MI patients at baseline TTO valuing current state Patientset al., 1991c 0.872–0.864 MI patients at 12-month versus full health

follow-up

Tsveat, 0.88 Treatment group TTO: 10 years in current Survivors of MI (n = 82)et al., 1995d 0.89 Placebo group state versus shorter life in

Valuing own health state excellent health

Nease, et al., 1.0 Angina class I TTO (VAS and SG Patients1995e 0.997 class II also available)

0.929 class III/IVValuing their own health

Kuntz, et al., 0.89 Mild angina/no CHF TTO: telephone survey Patients 1996f 0.85 Mild angina, CHF but no further details (1051 overall states)

0.82 Severe angina on procedure0.78 Severe angina, CHF

Valuing their own health

VAS, visual analogue scale; QWB, Quality of Well-Being [scale]; CHF, congestive heart failurea Grann, et al. J Clin Oncol 1998;16:979–85b De Haes, et al. Int J Cancer 1991;49:538–44c Oldridge, et al. Am J Cardiol 1991;67:1084–9d Tsevat, et al. J Am Coll Cardiol 1995;26:914–19e Nease, et al. JAMA 1995;273:1185–90f Kuntz, et al. Circulation 1996;94:957–65


64

from these studies to specific age groups of themodel would be to assume a constant absolutereduction regardless of age; another would be toassume a constant proportional effect on healthstate values. In the absence of good evidence, it is not possible to say which approach is correct.The latter approach was used in the model since it assumes that the better your health status themore you have to lose – which was considered to be the most realistic assumption.

Valuation techniqueAnother important difference between estimateswas the valuation technique used to elicit healthstate values, whether directly as part of the study orimplicitly through the use of generic preference-based measures. It is currently recommended thathealth state utility values should be obtained usinga choice-based technique such as SG or TTOrather than a rating scale.293 This recommendationwas used in selecting values for this model.

Anchor statesDifferent anchor states were also used by thestudies in their valuation tasks. For SG or TTO, ahealth state is valued against two reference states,one better and the other worse. Although moststudies used ‘dead’ as the worst state, differentupper states were described, such as ‘excellent’health, ‘full health’ (as defined by EQ-5D or HUI-II), ‘best imaginable for your age’, or ‘ yourcurrent’ health. This had important implicationsfor the interpretation of the estimates and theiruse in the model. By convention, death is given a value of zero and the upper anchor state is givena value of one. When a health state was valued in astudy using, for example, ‘best imaginable for yourage’ as the upper anchor, the values were higherthan those that would be generated from using‘full health’ as the upper anchor. In some studies,attempts were made to correct for this by trans-forming the results using valuations of the upperanchor, as in Gabriel and colleagues.288 The modelrequires a common scale, regardless of age; thusthe health state values must lie on a scale where a value of one is equivalent to full health.

Estimating the health loss from an eventThe model estimates the health loss for eachindividual from a health event as the differencebetween the health state values before and afterthe event. Earlier models assumed that the pre-event health state value was either one or theaverage health state value for the individual’s ageand gender group. However, this assumes thatthose who have experienced an event such as a hip fracture would have experienced full or

average health if the fracture had not taken place. Taking an age–gender matched sample from the general population as a control group at least corrects for the fact that most people whoexperience fracture, particularly hip fracture, areolder than the general population. According to a recent general population survey of the UK,for example, people in the age group 70–74 yearshave a health state value of about 0.8. A relatedapproach in a study by Oleksik and colleagues285

was to recruit a matched sample of people whohad not had a (vertebral) fracture and to use them as the control group.

These methods do not provide valid controlgroups for those avoiding a fracture, since thosewho have a fracture may have a different healthstatus from the average and are very unlikely tohave had a health state value of one prior to afracture. This concern was borne out in a recentstudy by Brazier and colleagues286 on patientsrecruited into a trial to reduce the rate of fracture.This study provided genuine prospective data on the impact of hip fracture on health. Healthstatus was assessed at entry into the trial prior tofractures occurring and then, following fracture, at 6 months and 12 months. Such data were notavailable for other health states in the model.

ReviewThe health state values considered in this revieware presented in Table 52, including their: healthstate descriptions, mean and SDs, the valuationtechnique employed and the source of thevaluations for each of the osteoporosis-relatedconditions. For comparison, normative health state utility values for the UK are presented by age group. These values were obtained from the results of the EQ-5D being administered toover 3000 representative members of the UKgeneral population. The values used by NOF are also presented for comparative purposes, since these are in common use in currenteconomic evaluations.

The 23 empirically derived health state utilityvalues for the four fracture conditions (i.e. hip,vertebral, wrist and established osteoporosis)differed considerably from the values obtained by a panel of experts for NOF. For example, theNOF value for vertebral fractures of 0.97 compareswith empirical values that range from 0.31 to 0.80. There was also a considerable range of values for each condition, probably due todifferences in the derivation of the estimates.Differences included what is being valued, thevaluation technique used, and the anchor states


65

used in the valuation task. The results arediscussed below in more detail for each condition.

Healthy osteoporosisThe model required a set of normative health state utility values. There were remarkably fewstudies with detailed normative data by age. In the USA, there was a set of normative data basedon a random survey of the population of one cityusing the Quality of Well-being scale and the TTOvaluation technique294 and, in Canada, a version of HUI was used in a state-wide health survey.295

The largest normative dataset of health state utility values available in the UK is based on theEQ-5D (administered to over 3000 representativemembers of the UK general population), and this was been used in the model.296

Established osteoporosisGabriel and colleagues298 reported on patients’own valuation of states diagnosed with estab-lished osteoporosis and obtained values similar to those that would be expected for their age. They also reported on the valuation of ahypothetical state of 0.43 using a sample of non-fracture cases. However, this value suffers from being based on a description that does notclearly relate to any one type of fracture andincludes references to future risk that are likely to distort the valuation of the state. Conceptually, a better approach would be to base the value of a case of established osteoporosis on the model estimate for the worst fracture experienced by each patient.

Hip fractureThere were nine different hip fracture valuesreported across three studies, ranging from 0.28 to 0.72. The HUI-II valuation of 0.68, forindividuals who had fractured their hips in the last 5 years, was significantly below the age/gender norm of 0.82 found in Canada,295 andsuggests a multiplier of around 0.83. However, in the study by Brazier and colleagues,296 betterdata were provided for use in the model as health state utility values were given for the sample population before they experiencedfracture, thus offering a more valid estimate of the loss in health status associated with a fracture.The mean health state utility values at 6 and 12 months after hip fracture were 0.49 and 0.48, respectively, compared with 0.6 at baseline.These figures imply a multiplier of 0.8.

Nursing homeThere was only one published estimate for hipfracture cases in a nursing home. Salkeld asked a group of elderly respondents to value a ‘bad’ hip fracture state that included ‘being in a nursing home’. However, the upper anchor used in the TTO question was a hypothetical state of someone in good health for his or her age and no account was taken of the likely health state of someone at most risk of fracture.Furthermore, the description of the health state was based on quantitative evidence and,hence, its relevance to the cases in the model is not clear.290 Hence, it was decided to use the NOF assumption of 0.4 in this model.


TABLE 52 Health state utility values used in the model

Health state Value Source

‘Healthy’ osteoporoticAge group (years) 45–49 0.840 Kind, et al., 1998296

50–54 0.85055–59 0.80260–64 0.82965–69 0.80670–74 0.74775–79 0.73180–85 0.69985+ 0.676

Established osteoporotic (use values associated with the type of fracture)Hip fracture 0.797 (95% CI, 0.651 to 1.012) Brazier, et al., 2000286

Nursing home 0.4 NoF9

Vertebral fracture 0.909 (95% CI, 0.84 to 0.97) Oleksik, et al., 2000285

Wrist fracture in first year 0.981 (95% CI, 0.978 to 0.986) Dolan, et al., 1999297

Proximal humerus 0.981 (95% CI, 0.978 to 0.986) Dolan, et al., 1999297

Breast cancer 0.62 (assumed range 0.33–0.84) Hutton, et al., 1996287

CHD 0.85 Assumption


66

Vertebral fractureThe empirical estimates for vertebral fracture wereconsiderably below the NOF assumption of 0.97.The lowest value was obtained from non-fracturerespondents for a hypothetical state of ‘multiple’fractures but this state is not used in the model.After allowing for the lower health state utilityvalues expected in those age groups prone to verte-bral fracture, the other apparent differences wereconsiderably reduced. The HUI-II estimate forthose who had a fracture in the last 5 years was 0.8,which compares with, for example, the normativevalue based on Canadian data of 0.82.a The bestdata for the model were provided by Oleksik andcolleagues,285 whose study was based on a sample ofclinically diagnosed vertebral fracture cases usingthe EQ-5D and had a larger sample of patients thanthe study by Gabriel and colleagues.288 The estimatefor individuals who had experienced a fracture inthe last 5 years compared with those who had notwas 0.75–0.82, which generated a multiplier of 0.91.The adequacy of the cross-sectional control was ofconcern in this study. The control group had tomeet the same inclusion criteria, including age and T score (< –2.5), but the authors found thatthe individuals in the control group were signifi-cantly younger (by 2.5 years), had a higher lumbarspinal BMD, and a lower prevalence of non-vertebral fractures. The consequences of thesedifferences for the EQ-5D score are not known.

Wrist fractureSome earlier models assumed that a wrist fracturehad no impact on health status. The one empiricalstudy in the field found a significant impact overshort periods.297 The researchers administered theEQ-5D at admission and at the final visit to theAccident and Emergency department, and wereable to estimate a mean loss in health state utilityvalue for the period since the wrist fracture byassuming a linear progression between the firstand last visits. Whether the EQ-5D would besensitive to some of the problems associated withwrist fracture was a cause for concern, particularlythe longer-term complications found in a smallproportion of patients. However, this is the onlyempirical estimate available at present.

Proximal humerusOn the advice of the clinical collaborators on this project, it was assumed that a fracture of the proximal humerus had the same impact onhealth status as a wrist fracture.

Breast cancerThere have been a number of studies presentingempirically derived estimates of the impact ofbreast cancer on health state utility values. Theutility value depends on whether the disease isstable or progressive, whether it is being treated,and its stage. There was no average value for thisdisease and, hence, it was necessary to select avalue that broadly represented the consequencesof breast cancer for a person’s health status. It was decided to use the value for stable diseaseestimated by Hutton and colleagues.287

CHDThis disease also suffers from the problem of being associated with more than one condition.There were estimates for patients following MI and other values for different severities of angina.An assumption of 0.85 was used in the model, with a range of 0.72–0.99 depending on the type of CHD and the study method.

A reference case set of values for the modelThere were wide ranges of preference-based health state utility values for each condition,primarily because of differences in the descriptivesystems and the sample of respondents used in thevaluations. One recommended solution to such asituation is for all analysts to use a reference caseof values. This does not imply that analysts shouldonly use the reference case in any future economicevaluation but that they should be used in at leastone analysis of each economic evaluation of anintervention for osteoporosis.

The influential Washington Panel on Cost-Effectiveness recommends the use of a genericinstrument with social valuations of health statesobtained using a preference-based instrument.298

This allows comparison between healthcare pro-grammes, such cardiac failure or cancer versusosteoporosis, as well as within programmes. The problem to date with the condition-specificapproach has been that this has been limited toone or two vignettes, and these do not necessarilyreflect the full range of states associated with eachcondition. Furthermore, they cannot be linkedeasily to patients in trials. Generic instruments can be administered to patients in trials or otherclinical studies and thus provide a more accuratequantitative basis to the descriptive results. Whileaccepting that there may be problems with generic

a See: Roberge R, Berthelot J-M, Cranswick K. Adjusting life expectancy to account for disability in the population: a comparison of three techniques. Social Indicators Res 1999;48:217–43.


67

health state classifications for some conditions, such as insensitivity to the consequences of wristfracture, another approach would be to produce a preference-weighted condition-specific measure.

In this review, two generic preference-basedmeasures were found to be in use, EQ-5D andHUI-II. There are few data on their relativeperformance in osteoporosis and no methodo-logical basis for preferring one to the other.Currently, the EQ-5D has the advantage of beingavailable for more osteoporosis-related conditionsthan the HUI-II and, hence, is preferred for thereference case set of values. Because of lack of evidence, it was not possible to distinguish between first and subsequent years, unlike NOF.

The final selection of health state utility valuesused in the model are presented in Table 52,including the mean ‘multiplier’ and 95% CIsestimated from the studies using Feiller’s theorem.There are many uncertainties surrounding theappropriate estimates to use in the model for thereasons given above; hence it was important for afull sensitivity analysis to be undertaken, using 95%CIs to examine the robustness of any conclusionsdrawn to the health state utility values used.

Research agendaRemarkably few studies were found relating to the impact of osteoporosis-related conditions onhealth state utility values. This finding was con-firmed by a recently published listing of 1000health state utility values,299 which contained only six values for these conditions, five for hip and onefor vertebral fracture, all of which were based onexpert opinion rather than empirical evidence. Thestudies reviewed here had begun to use acceptedmethods for use in economic evaluation but werelimited in terms of age range, sample size, theperiod since the event and poor control groups. To improve the reference-case value dataset wouldrequire the administration of a preference-basedgeneric health status measure to a large prospectivepopulation cohort and long-term follow-up. Suchpreference-based measures could include the EQ-5D, HUI-III or the recently developed ShortForm (SF)-6D that utilises SF-36 data. The choiceshould depend on evidence of their validity acrossthese conditions. It would be possible and import-ant to estimate by age the actual loss in healthstatus utility values over time following each of the fractures (including multiple fractures) and to generate measures of variance. These data could be collected both as a part of large clinicaltrials and observationally. International studieswould also allow for cross-national comparisons.

A longer-term agenda should look more critically at the instruments used for estimating health status utility values and, in particular, the genericpreference-based measures for each of the fractures.If generic measures were found to be irrelevant orinsensitive to important aspects of one or more ofthe conditions, then another approach would be to develop condition-specific vignettes, althoughthese are difficult to apply to quantitative data fromtrials and other studies. A third approach would be to develop condition-specific, preference-basedmeasures for use on patients in clinical studies (thesecould use existing measures of health-related qualityof life). A fourth approach would be to estimatepreference weights for condition-specific measures.

A review of costing

The model developed for this study required costestimates for the health states associated withestablished osteoporosis, the treatment costs forthe drugs under review, and the adverse drugreactions from these treatments. Costing wasrequired for the following health states:

• hip fracture• confinement to nursing home due to hip fracture• death due to hip fracture• vertebral fracture• wrist fracture• other fractures• breast cancer• death due to breast cancer• CHD• death due to CHD• additional ‘healthy’ life-years.

The model also required that costs be disaggre-gated into year of incidence and subsequent yearcosts, and that costs be weighted for age. Poten-tially, the model required costs to be estimated for upwards of 120 variables. Ideally, they would be estimated using prospective resource use datacollected alongside an appropriate UK-basedrandomised trial. Given the absence of such data,the method of deriving the best available costestimates for use in the model is described below.

MethodsDrug formulation costs were estimated usingpublished UK reference prices from the BritishNational Formulary 2000 (no. 39). It was assumed there were no administration costs.

The published literature was searched for costs or information that would help in costing the



68

modelled health states. The review drew on papersidentified from a series of systematic searches,including searches of papers reporting economicevaluations, using electronic databases includingMEDLINE, NHS EED, HEED, EMBASE, and theScience Citation Index. Economics studies wereidentified using a search filter based on thatdeveloped by the NHS Centre for Reviews andDissemination.300 Details of the searching andselection strategy are outlined below and the searchterms are presented in appendix 6. Generally,priority was given to the most recent UK-basedanalyses, as treatment pathways and unit costs candiffer significantly between countries and over time.

The initial search resulted in over 2000 titles beingreturned. These were filtered by title and then ab-stract, in order to exclude irrelevant papers. Theinitial filtering by title reduced the number of papersto 200. In addition to the abstracts of those papersbeing reviewed, any non-UK papers plus those pub-lished before 1990 were set aside. Papers not ex-plicitly addressing costing but analysing key resource-use variables, such as inpatient length of stay (LOS),were included. The resulting 29 papers were re-quested and reviewed. The costing analyses presentedin the short-listed papers were assessed for quality, toidentify whether the papers presented appropriatelevels of detail for selected criteria (Table 53).2,3,11,20,216,

217,284,301–322 The results of the selection and qualityassessment process were used to identify the mostappropriate papers for the modelling exercise.

The very specific cost data requirements of the modelmeant that data gathered from the searched literaturewould only partially populate the model. In particular,information on age-related costs and the division ofcosts between year of incidence and subsequent yearsproved to be sparse. Also, because the economicsliterature search was specifically targeted at osteo-porosis and related fractures, it was not a good sourceof information for costing adverse drug events, suchas breast cancer and CHD. When costing informationwas not available from the literature, other sources,including publications from ad hoc searches andpapers known to the authors, were used.

In view of the lack of information on age-relatedcosts, a local patient database that included infor-mation on age, diagnosis codes, destination ondischarge, and LOS was analysed. Inpatient LOShas been assumed to be a good proxy for resourceuse and has been used to determine the appro-priateness of age weighting for modelled healthstate costs. In brief, age-specific, mean patient LOSwas calculated by health state, allowing for desti-nation on discharge (including death). Student

t-tests were then performed. When the statisticalanalysis indicated evidence for age weighting, theliterature was searched for appropriate age weights.If no age weights could be found, the ratio of theage-specific to all-age mean LOS, calculated usingthe local data, was used as a proxy age weighting.All age weighting estimates are presented in theappropriate sections of this report.

ResultsPublished literatureThe searched papers predominantly reportedcosting analyses for osteoporotic fractures. Of the29 papers, nine exclusively examined hip fracturecosts. Only three papers included references toCHD and breast cancer, and none of these provedto be helpful in populating the model. The finalselection of articles included six review articles.Only a handful of the papers fully satisfied thequality assessment criteria, although these criteriawere not wholly appropriate for the review articlesand those papers in which resource use (e.g. LOS)rather than costs were analysed. In summary, the published literature proved somewhat limitedin terms of quantity, quality, and appropriatenessfor the model. That said, the paper by Dolan and Torgerson20 was a key source for estimatingcosts for all osteoporotic fractures.

Cost estimatesWhen deriving the cost estimates for the model,the costing took, where appropriate, an NHS andsocial care perspective. Patient costs and indirectcosts to the economy were not considered. In theanalyses presented below, the costs quoted use thefinancial year currencies presented in the originalsources. For modelling purposes, the derived costsestimates have all been inflated to 1999/2000financial year prices using the Hospital andCommunity Health Service (HCHS) pay and prices index,323 supplemented by the GDP [grossdomestic product] deflator for years 1998/89 and1999/2000.324 The various costs estimates used inthe model are presented in Table 54.

Drug costsThe estimated annual drug costs for the formu-lations under review in this report are presented in Table 55. The table shows the wide choice ofHRT formulations on offer in the UK. Those drugs used in the model are indicated in bold type.Fluoride is not commercially available in the UKbut is used by specialist centres. The cost used was the pharmacy price in France for Osteofluor®

(Merck Pharmaceuticals), 50 mg daily (1.41 FF),equivalent to £48 per year in current prices usingan exchange rate of £1 = 10.69 FF.


69


TABLE 53 Summary analysis of costing papers found in the literature search

Study Procedures/ Resource Methods of Methods of Year of Sensitivity Discount Summary notesdrugs quantities estimating estimating prices analysis rate analysed identified quantities prices given? given? performed? used

separately given?from prices

Beech, et al., Hip fracture N/A N/A N/A N/A N/A N/A Good LOS data by type1995301 of hip fracture – see

Withey.322 Also quotedsome orthopaedic costs/day

Best & Hip fracture No No No Some Yes 6% Some local treatment Milne, 1998302 costs for Wessex

Daly, et al., None No N/A Yes 1989/90 Yes 6% Potentially useful 19922 modelling paper for HRT

and LYG but not muchon costs

Daly, et al., All fractures No No Yes 1992 Limited 6% Expected cost and savings 1996217 and CHD of HRT and treating

side-effects

Dolan & All fractures Yes Yes Yes 1995/96 Limited N/A Good study. Data and Torgerson, methods rich for fractures199820

Drummond, CHD No No No N/A Yes N/A Methodological discussion et al., 1993303 paper on choice of

outcome measures for drug therapy in hypercholesterolemia

Fox, et al., Hip fracture Some LOS Yes No N/A No N/A Hip fracture LOS and 1993304 location on discharge for

two UK units

Francis, et al., Vertebral Yes Yes Yes 1994? No 6% Focus on costs of drugs1995305 fracture for vertebral fracture

French, et al., Hip fracture Only LOS Yes Yes 1993 No None Detailed costs for hip 1995306 fracture – regression

analysis

Hollingworth, Hip fracture Yes Yes Yes 1991/92 Limited N/A Hip fracture, age/sex et al., 1995307 resource use data

Hollingworth, Hip fracture Yes Yes Yes 1991/92 Yes N/A Focus on potential savings et al., 1993308 from early discharge

Morris, et al., HRT, No Yes Yes 1998 Yes 6% Modelling paper with focus 1999309 raloxifene, on cost-effectiveness

hip fracture, analysis (LYG) of HRT and CHD, breast raloxifene; no primary data cancer other than drug costs

O’Cathain, Hip fracture Only LOS YES Yes 1992 No N/A Focus on potential savings 1994310 from hospital at home

scheme

Parker, et al., Hip fracture, N/A N/A N/A N/A N/A N/A Analysis of variations in 1998311 LOS LOS after hip fracture

Pitt, et al., Hip and No N/A Yes 1989 Yes 6% Costs of HRT and likely im-19903 spine fracture pact on fracture incidence

Reid & Etridonate Review Review Review Review Review Review Review paper with focus on Torgerson, likely cost-effectiveness of 1998312 etridonate compared with

other drugs. Notes lack of good quality costing data.Quotes some indicative drug and hip fracture costs

continued


70

TABLE 53 contd Summary analysis of costing papers found in the literature search



Sculpher, Prevention Review Review Review Review Review Review Critical structured review,et al., 1999284 and drug costs presented,

treatment of excludes costing studies,osteoporosis suggests US prices

inappropriate

Taylor & Cardiac Yes Yes Yes Yes Yes 5%? Concentrates on cardiac Kirby, 1999313 rehabilitation rehabilitation but contains

potentially useful more general cardiac costing information and sources

Torgerson & Osteoporotic N/A N/A N/A N/A N/A N/A Prescribing after Dolan, 1998314 fracture osteoporotic fractures.

No costing data. Some sources for drug papers

Torgerson & Oral and Yes Yes Yes No None 6% Measures direct costs of hip Kanis, 199511 injected fracture using Hollingworth

vitamin D, results. Possible wrong hip fracture decision criteria, i.e. average

not marginal. Implies that addition of indirect costs would double costs

Torgerson & HRT Review Review Review Review Review Review Review article including Reid, 1999315 cost of HRT

Torgerson & Screening for Review Review Review Review Review Review Focus on cost-effectiveness Reid, 1993316 prevention of of screening for prevention

osteoporosis of osteoporosis and hip fracture. No primary data;some assumptions and use of published data. Useful general discussion

Torgerson & Osteoporosis Review Review Review Review Review Review Review: no detailed costings Reid, 1997216 but good summary of

costing methodologies and quality-of-life methods

Torgerson, Drug costs Yes Yes Yes No Yes 6% Case study in use of et al., 1996317 and fracture economics for prioritising

avoidance research using cost-effectiveness of drugs to prevent hip fracture.No primary data

Torgerson, BMD No Yes No No N/A Not Discussion of value of BMD et al., 1996318 stated scans. Some drug and BMD

scan costs

Torgerson, Fracture Review Review Review Review Review Review Economics of prevention et al., 1997319 prevention of osteoporosis. Potentially

useful summary table of previously published economic analyses

Townsend, HRT Yes Yes Yes 1994 Yes N/A HRT cost trends. Considers 1998320 total costs but some

potentially useful break-down of types of HRT use in UK

Townsend & HRT No Yes Yes 1992 No Not Cost-effectiveness scenario Buxton, 1997321 stated analysis for proposed trial

of HRT

continued


71

Fracture costsDolan and Torgerson20 presented detailed costingsof hip, vertebral, wrist and ‘other fractures’ in theirpaper. They estimated costs by analysing resourceuse in terms of acute care, social and long-termhospital care, follow-up and drug use.

Hip fracturesDolan and Torgerson20 presented detailed costingsfor various categories of hip fracture patients andfor various stages of treatment from acute to long-term domestic and residential home treatment. Formodelling purposes, cost estimates needed to be

constructed for hip fracture patients dischargedhome, hip fracture patients discharged to long-term residential or hospital care, and for hip fracture patients dying within the year of fracture. An acute care cost of £4808 per patient was estimated. This estimate has been included in all three of the model’s hip fracture categories.

Dolan and Torgerson20 also presented compre-hensive estimates for social care costs, including an estimate of £1574 for care in the homefollowing discharge from hospital. This estimate


TABLE 53 contd Summary analysis of costing papers found in the literature search



Withey, Hip fracture N/A N/A N/A N/A N/A N/A No costings but potentially et al., 1995322 useful resource use

information for costing of hip fracture by case-mix.Same data and team as Beech301

LYG, life-years gained

TABLE 54 Final cost estimates by health state, age and time (year 1999/2000 prices)

Health state Costs per annum (£)

Base case Age range (years)

45–64 65–74 75–84 85+

Year 1 Year 2 Year 1 Year 2 Year 1 Year 2 Year 1 Year 2 Year 1 Year 2

Suffered uncomplicated 7,398 N/A 4,530 N/A 5,698 N/A 7,499 N/A 8,600 N/Ahip fracture

Confined to nursing 30,360 21,163 27,491 20,695 28,659 21,291 30,460 22,179 31,562 23,362home due to hip fracture

Death due to hip fracture 8,666 N/A 8,666 N/A 8,666 N/A 8,666 N/A 8,666 N/A

Suffered vertebral fracture 465 216 419 195 521 242 546 253 590 274

Suffered wrist fracture 522 N/A 316 N/A 316 N/A 516 N/A 1,609 N/A

Suffered ‘other’ fracture 1,492 N/A 902 N/A 902 N/A 1,474 N/A 4,601 N/A

Contracted breast cancer 7,949 N/A 7,949 N/A 7,949 N/A 7,949 N/A 7,949 N/A

Death due to breast 10,981 N/A 10,981 N/A 10,981 N/A 10,981 N/A 10,981 N/Acancer

Contracted CHD 2,577 619 1,915 619 2,454 619 2,930 619 3,360 619

Death due to CHD 2,160 N/A 2,160 N/A 2,160 N/A 2,160 N/A 2,160 N/A

Confinement to nursing 21,163 21,163 20,695 20,695 21,291 21,291 22,179 22,179 23,362 23,362home not due to hip fracture

Healthy life-years 776 N/A 416 N/A 874 N/A 1,558 N/A 2,468 N/A


72

TABLE 55 Drug costs (entries in bold type are those used in the model)

Drug class Drug name Dosage Preparation Unit cost Annual cost(£) (£)

SERM Raloxifene® a 60 mg 28-tablet pack 19.76 257

Calcitonin Calsynar® b 100 unit/ml 1-ml ampoule 6.34 2,314

Miacalcic® c 100 unit/ml 1-ml ampoule 7.13 2,602

Bisphosphonate Fosamax® d 28-tablet pack 25.69 334

Didronel PMO® e 400 mg 90-tablet pack 40.20 163

Calcium supplements See vitamin D

Calcium salts Calcichew® f 1.25 mg 100-tablet pack 10.97 40

Calcichew Forte® f 2.5 mg 100-tablet pack 21.94 80

Parenteral Gluconate 8.9 mg 10-ml ampoule 0.57preparations injection 10%

Calcium chloride 100 mg/ml 10-ml disposable 4.02injection syringe

Vitamin D Adcal-D3® g CaCo 1.5 mg 100-tablet pack 7.50 55

Calcichew D3® f 1.25 mg, 200 units 100-tablet pack 13.65 100

Calcichew D3 Forte® f 100-tablet pack 16.50 120

Calciferol 10,000 units 20-tablet pack 4.65(1.25 mg (50,000 units) may also be available)

Calciferol 7.5 mg 1-ml ampoule 5.92 7.07injection (300,000 units)/ml 2-ml ampoule

Calcitriol Calcijex® h Injection, 1 µg/ml 1-ml ampoule 5.71

Injection, 2 µg/ml 1-ml ampoule 11.42

Rocaltrol® i 250 ng 20-tablet pack 4.31 157

HRT for women without a uterusConjugated Premarin® j 6.25 µg 3 x 28 tablets 7.36 32oestrogens 1.25 mg 3 x 28 tablets 9.99 43

Estradiol only Organon implants 25 mg per implant 9.59 1950 mg per implant 19.16 38100 mg per implant 33.40 67

Elleste-solo® k 2 mg tablets 3 x 28 pack 5.34 23Elleste-solo MX 80® k 80 µg/24 hours 8-pack (two per week) 6.56 85Estraderm MX50® c 50 µg/24 hours 8-pack (two per week) 7.45 97Estraderm MX75® c 75 µg/24 hours 8-pack (two per week) 7.90 103Estraderm TTS50® c 50 µg/24 hours 8-pack (two per week) 7.45 97Evorel 50® patchesl 50 µg/24 hours 8-pack (two per week) 7.45 97Evorel 75® patchesl 75 µg/24 hours 8-pack (two per week) 7.90 103Evorel 100® patchesl 100 µg/24 hours 8-pack (two per week) 8.20 107Fematrix® m 80 µg/24 hours 8-pack (two per week) 6.95 90Femseven 50® d 50 µg/24 hours 4-pack (one per week) 6.44 84Femseven 75® d 75 µg/24 hours 4-pack (one per week) 7.49 97Femseven 100® d 100 µg/24 hours 4-pack (one per week) 8.19 106Oestrogel® n 1.5 mg 64-dose pack 7.95 45Progynova® o 2 mg 3 x 28-pack 7.02 30Zumenon® m 2 mg 84-pack 7.65 33

continued


73

was derived from a synthesis of patient survey data and other published sources.325,326 The totalannual social care cost for hip fracture patients was estimated to be £370.9 million. This figure has been apportioned across the model’s three hip fracture patient groups by assuming that‘discharged to home’ patients form one patientgroup, that the two patient groups dying in theyear following fracture form a second patientgroup, and that patients assumed to remain inlong-term hospital care or residential care afterdischarge form a third group. Thus the social care costs per patient for the three defined groups for the year immediately followingtreatment are estimated to be as follows.

Patients discharged home (45%) £1,574Patients dying within the year (30%) £2,964Long-term care patients (25%) £22,218

The total cost estimate of £8.5 million for addi-tional general practitioner (GP) and follow-upoutpatient visits estimated by Dolan and

Torgerson20 have been apportioned to all patients discharged from hospital care within the first 12 months after fracture. Thus GP andoutpatient average annual costs are estimated to be as follows.

Patients discharged home £253Patients dying within the year £0Long-term care patients £202

Aggregating these hip fracture cost estimates, gives year-of-fracture cost estimates per patient as follows.

Patients discharged home £6,635Patients dying within the year £7,772Long-term care patients £27,228

These figures represent the year one cost estimates for the defined hip fracture patientgroups, ‘suffered an uncomplicated hip fracture’,‘died from hip fracture’ and ‘confinement tonursing home due hip fracture’, respectively.


TABLE 55 contd Drug costs (entries in bold type are those used in the model)

Drug class Drug name Dosage Preparation Unit cost Annual cost(£) (£)

HRT for women without a uterus contdEstradiol, estriol Hormonin® f 600 µg 90-tablet pack 6.44 26andestrone only Hormonin® f 600 µg 90-tablet pack 6.44 52Estropipate Harmogen® f 1.5 mg 28-tablet pack 3.14 41Oestrogens and HRT Tibolone 2.5 mg daily 28-tablet pack 13.66 178

HRT for women with uterusConjugated Premique® j 3 x 28-pack 22.62 98oestrogens Prempak-C® j 625 µg 3 x 40-pack 13.38 58Estradiol with Climesse® c 2 mg 28-pack 7.90 103progestogen Cyclo-progynova® p 2 mg 28-day pack 3.50 46

Elleste-Duet® k 2 mg 3 x 28-day pack 9.72 42Estracombi® c 50 µg/24 hours 8-patch pack 11.14 145Estrapak® c 50 µg/24 hours One-month pack 9.48 123Evorel conti patch® l 50 µg/24 hours 8-patch pack 12.90 168Evorel pak® l 50 µg/24 hours 8-patch pack 8.45 110Femapak80® m 80 µg/24 hours 8-patch pack 8.95 116Femoston 2/10® m Estradiol, 2 mg 28-tablet pack 4.99 65Femoston 2/20® m Estradiol, 2 mg 28-tablet pack 7.48 97Kliofem® q 2 mg 3 x 28-tablet pack 25.95 112Nuvelle® o 2 mg 3 x 28-tablet pack 13.77 60Tridesta® r 2 mg 91-tablet pack 24.90 100Trisequens® q 2 mg 3 x 28-tablet pack 20.55 89

a Eli Lilly & Company; b Rhone-Poulenc Rorer; c Novartis Pharmaceuticals; d Merck Sharp & Dohme; e Procter & Gamble Pharma-ceuticals; f Shire Pharmaceuticals; g Straken; h Abbott Laboratories; i Roche Products; j Wyeth Laboratories; k Searle; l Janssen-Cilag;m Solvay Healthcare; n Hoechst-Marion Roussel; o Schering Health Care; p ASTA Medica; q Novo Nordisk Pharmaceuticals;r Orion Pharma


74

Costs for subsequent years were not estimated byDolan and Torgerson,20 and very little informationto help in the estimation of these was found in thepublished literature. Consequently, a number ofsimplifying assumptions were made. Given that all patients who died of their hip fracture wereassumed to die within the year of fracture,subsequent year costs were not an issue for thispatient group. It was also assumed that patients inthe ‘uncomplicated hip fracture’ group incurredno costs beyond the year of fracture. A possiblejustification for this simplifying assumption is the prescribing work by Torgerson and Dolan,314

which indicated a return to control levels ofprescribing for hip fracture patients in the year following the year of fracture.

In the analysis by Dolan and Torgerson,20 all long-stay hospital and residential care patients were assumed to have been discharged to privatenursing homes for the second and subsequentyears following their fracture. Consequently, theywere assumed to incur subsequent year costs of£18,980 per annum, with no additional costs forhip fracture care. This figure has also been used as the cost estimate for any patient confined to a nursing home in the model.

Analysis of local LOS data indicated some justifi-cation for age-weighting costs for hip fracturepatients discharged home but not for those whodied following hip fracture. On the basis of thelatter, no age weights were applied to this patientgroup. Hollingworth and colleagues307 presentedsecondary care hip fracture costs by age andgender. Using their results for female hip fracturepatients, the analysis by Hollingworth andcolleagues implied the following hip fracture age weightings:

• age 45–64 years, 0.61• age 65–74 years, 0.77• age 75–84 years, 1.01• age 85+ years, 1.16.

These weights were applied to the year 1 costestimates for the ‘uncomplicated hip fracture’patient group in the model.

For the group of hip fracture patients dischargedto residential care, it was assumed that the costs of medical care were the same in year 1 as forpatients discharged home (£6635), and that thiscost element for this patient group had the sameHollingworth age weightings.307 The remainder of the £27,228 was assumed to be invariable by age.

The final group of hip fracture patients for whomage weightings needed to be considered were those living in nursing homes beyond the year of fracture. Dolan and Torgerson20 estimated costsfor nursing home care using data from Netten and Dennett,326 who indicated that 5% of privatenursing home costs comprised fees for externalservices such as district health authority and GPservices. On this basis, it was assumed that 5% of the estimated cost should vary by age. Theweightings applied were determined using HCHSper capita expenditure figures (see below fordetails).327 The model used the same age weight-ings as for those patients confined to nursinghomes for reasons other than hip fracture.

Wrist fractureDolan and Torgerson estimated the cost per wrist(Colles) fracture as £468 per annum.20 This wascomprised of £368 for acute care costs, £64 for GPcosts and £36 for outpatient costs. This figure hasbeen used for the year-of-incidence cost estimatefor wrist fracture in the model. Because of thenature of wrist fractures, it was assumed that allcosts associated with wrist fractures are consumedin the year of fracture; hence, there are nosubsequent year costs. Statistical analysis of localpatient data indicated that there were significantdifferences in LOS for the different age groupswho sustained a Colles fracture. These data wereused to derive the following age weightings:

• age 45–74 years, 0.60• age 75–84 years, 0.99• age 85+ years, 3.08.

Vertebral fractureDe Laet and colleagues acknowledged that costinginformation for vertebral fractures was sparse.328

Although short-term costs are not large comparedwith, for example, hip fracture, vertebral fracturesare more chronic in nature. Despite this, thepublished evidence for long-term resource use is poor.

Dolan and Torgerson estimated an annual cost of £479 (1995/96) per patient for the year offracture.20 This included £62 for costs of bonedrugs administered after the fracture. Becausecosts of bone drugs are modelled separately in the current report, the £62 was deducted to avoid double counting. Vertebral fracture costswere therefore assumed to be £417 per patient in the first year.

Given the chronic nature of vertebral fractures,subsequent year costs will be relatively more


75

significant than the other fractures modelled.Dolan and Torgerson did not explicitly presentlonger-term costs for vertebral fractures;20

however, in a separate paper, Torgerson and Dolan reported that vertebral fracture patients(unlike hip and wrist fracture patients) showed a significant increase in prescribing of bone drugsin the year subsequent to the year of fracture.314

Given that such prescribing is likely to be accom-panied by additional GP and outpatient visits,subsequent year resource use from vertebral frac-tures is reasonably assumed. Acute costs associatedwith vertebral fracture were assumed to be con-fined to the year of fracture. Looking beyond UK-based evidence, the Dutch-based analysis by DeLaet and colleagues328 indicated that, comparedwith a control group, costs of visits to physicians fell by 40% in the year following a first vertebralfracture. Consequently, it was assumed that 60%(£193) of Dolan and Torgerson’s20 estimated costsof GP and outpatient follow-up visits persist assubsequent-year costs in the model.

None of the reviewed UK-based papers indicatedthe appropriateness of age weighting for vertebralfractures. In an analysis based in the USA, Chris-chilles and colleagues modelled the costs ofvertebral fracture for defined age groups.329

The results of this analysis imply weightings of0.90, 1.12, 1.17 and 1.27, respectively, for our fourage groups. These weights were applied to both year 1 and subsequent year costs in the model.

Other fracturesDolan and Torgerson20 quoted an acute cost of£1200 for ‘other’ fractures, using closed upperlimb fracture as a proxy for ‘other’ fractures. GPand outpatient follow-up costs, including rib,humerus and leg fracture data, were estimated at£138 per patient per annum. The estimated totalcost per patient was, therefore, £1338. The ageweighting and subsequent year assumptions usedfor Colles’ fracture were also applied to theestimate for ‘other’ fracture costs.

Adverse drug reactionsBreast cancerBased on earlier work by Wolstenholme andcolleagues,330 Dolan and colleagues presented a detailed costing analysis that covered all aspectsof treatment for breast cancer in the UK.331

Using 1995/96 prices, costs were analysed usingthe four recognised stages of cancer development.Individual estimates of treatment costs wererequired by the model for patients who survivedand those that died from cancer. In the model, the simplifying assumption was made that the

costs identified by Dolan and colleagues331

for Stages I–III comprised the costs for cancersurvivors. Stage IV costs formed the basis forcosting death from breast cancer.

Dolan and colleagues reported individual costs forsecondary, primary, hospice and nursing residentialcare.331 Their secondary care cost analysis alsoincluded estimated costs for breast reconstructionsurgery. These costs (£2046 per patient) wereapportioned to the survivor group in the model.

The costs of additional GP visits for breast cancer patients were estimated to be £141 perpatient by Dolan and colleagues.331 These costswere apportioned across both defined patientgroups. The costs of hospice and nursingresidential care were estimated by Dolan andcolleagues at £1.1 million for 443 terminally illpatients.331 This cost was apportioned across all1456 Stage IV patients in our analysis, implying a mean cost of £781 per patient. Thus, beforeallowing for inflation, cost estimates for breastcancer patients of £7129 for survivors and £9848 for those who died were used in the model.

According to Dolan and colleagues,331 theincremental costs of breast cancer tended to occur in the 2 years following onset of disease. On the basis of this evidence, the simplifyingassumption that all costs for breast cancer occur inthe year of incidence was made. Local LOS dataanalysis implied no need for age weighting for thecosts of surviving or dying from breast cancer.

CHDThe costing assumptions for CHD were based on the work of Pickin and colleagues332 and Piercy and Pledger.333 Pickin and colleaguesestimated secondary care costs for major CHDevents: that is, coronary artery bypass graft £5500;percutaneous transluminal coronary angiography£3517; emergency MI £1887; emergency ischaemicheart disease £1471 (1995/96 prices assumed).Based on the distribution of such events for a UK population, as quoted by Piercy and Pledger,333

a weighted average cost of £1937 has been calcu-lated for major CHD events. Increasing this by 5%to allow for outpatient events333 gave an estimatedaverage total cost for CHD of £2034. In addition, it was assumed that CHD patients consumed 6 months’ supply of simvastatin in the year ofincidence (27.4 mg daily @ £1.52 = £277.40),thereby implying an estimated average total cost of £2311 for year 1. It was assumed in the modelthat CHD patients would consume simvastatin for



76

the remainder of their lives, implying annualsubsequent year costs of £555 (1995/96 prices).

Estimates of the costs for patients who died from CHD were also required. The simplifyingassumption was made that patients who died in the year of a CHD event did so at the end of theacute inpatient episode. Consequently, a cost of£1937 (1995/96 prices) was estimated for deathfrom CHD.

Statistical analysis of local patient data implied that LOS did vary with age for patients dischargedhome but not for patients dying from CHD. Assuch, no age weightings were applied to thosepatients dying from CHD. Using the results of the LOS analysis, the following age weightings were assumed:

• age 45–64 years, 0.60• age 65–74 years, 0.60• age 75–84 years, 0.99• age 85+ years, 3.08.

Given that it is assumed in the model thatsubsequent year costs for CHD patients comprisedonly the costs of simvastatin, no age weightingswere assumed for these costs.

Healthy life-yearsThere is an option in the model to consider the costs of healthy life-years for the modelledpopulation. Also, some modelled patient groups(e.g. those discharged to residential care) haverequired age-related cost weightings. Age-relatedcosts for this population have been estimated using general health expenditure data. In theOffice of Health Economics compendium ofhealth statistics,327 the following age-related percapita expenditure (1996/97 prices) for HCHS for England are quoted:

• age 45–64 years, £383• age 65–74 years, £805• age 75–84 years, £1435• age 85+ years, £2274.

Using mid-1996 age group population estimates for England,334 the weighted average per capitaexpenditure for all individuals over 44 years of agehas been estimated at £715. As such, the followingcost weightings were estimated for the HCHS percapita expenditure figures:

• age 45–64 years, 0.54• age 65–74 years, 1.13

• age 75–84 years, 2.01• age 85+ years, 3.18.

The model also contains assumptions about GP visits and BMD scanning activity for patientswho were alive and being treated for osteo-porosis. Netten and Curtis indicated costs per GP consultation ranging from £13 to £18(1999/2000).335 Torgerson and colleagues318

quoted a BMD scan cost of £25 from a paper by Garton and colleagues,336 which quoted total scanning costs varying from £21 to £25.Precise timings of costings were not made explicit in the paper, hence, 1991/92 prices were assumed. Applying an HCHS inflator, theestimated 1999/2000 cost was £27.61–32.87. Based on these findings, costs of £16 per GPconsultation and £30 per BMD scan were used in the model.

Uncertainties surrounding cost estimatesTaking account of the model’s requirements that health state costs were estimated by age and broken down by year of incidence andsubsequent-year costs, the analysis presented here required estimation of well over 50 healthstates costs. Inevitably, these estimates were subject to uncertainty around the central estimates. The drug costs presented in Table 55include a range of formulations and, hence, cost estimates. The age- and condition-related cost variations for health states (Table 54) could be used as a basis for sensitivity cost estimates for the modelled health states. It is unfortunatethat the papers used as a basis for calculating the cost estimates used here presented very limited or, more usually, no sensitivity analyses. In view of this, and in view of the considerabletime implications of having to undertake such an analysis, further estimations of costing sensi-tivity ranges were considered to be beyond thescope of this study. The model was, however, used to test the sensitivity of cost-effectiveness to changes in drug prices.

SummaryThe costing data requirements of the modelstretched the available literature to their limits.Only a handful of papers proved to be of use interms of providing information for estimating costs for modelled health states. In particular,evidence for longitudinal and age-related costs was sparse. In view of the lack of publishedevidence, the authors were forced to take apragmatic approach to costing by making the best use of the available information.


77

Information on effectiveness of interventions(chapter 2), and the risk functions, health states

and costs (chapter 4) have been used to populate acost–utility health economics model. The prin-ciples of the model and its inherent assumptionsare described below.

Model approach

The approaches used previously were based oncohort analyses using decision-analysis and Markovmodels.337,338 The present model is an individualpatient-based, transition state, osteoporosis modelcreated in Excel 97© (Microsoft Corporation). Inthis type of approach, patients are modelled asindividuals and whether or not an event will occurin the forthcoming year is simulated for eachpatient. A full patient history is recorded and thus factors such as prior fractures and currentresidential status can be used to determine thelikelihood of events in the next period. Following a simulated event, the quality of life of the patientand the costs incurred are calculated. Any residualcosts of quality-of-life impacts from previousfractures are taken into account for both thesefactors. The model continues to simulate at 1-yearintervals until either the patient dies or the user-defined analysis period (e.g. 10 years) is reached.The process is repeated until all patients have beensimulated. The rationale for using the individualpatient approach is that it provides more accuracyand flexibility than a cohort approach, which isbounded by a limited number of transition states.Some examples are reviewed briefly below.

The first example relates to the accuracy withwhich the probability of fractures can be calcu-lated, based on the patient’s history. There is abreadth of published literature that indicates that an initial fracture greatly increases the risk of subsequent fractures263 (see chapter 4). Giventhis, it would be inaccurate to structure a cohort-based transition state model. Consider the exampleof two identical patients with osteoporosis at thecohort model initiation, for whom 5 years of lifeare simulated. Patient A may suffer no fractures in the first 4 years and a wrist fracture in the fifthyear. Patient B suffers no fractures in the first 2 years and then suffers a hip, vertebral and wrist

fracture in the next 3 years. In a simple cohortmodel, both patients would now reside in the wrist fracture state. However, if the values from the available data are used, patient B would be at much greater risk of vertebral fracture and at increased risk of hip fracture compared withpatient A. Without adjusting for this increasedprobability of fracture, the model would under-estimate the number of fractures that occurred.

As a further example, a large component of costs are those associated with nursing home care following hip fracture. If the model does nottrack the residential status of a patient, there is aprobability that additional nursing home costswould be added for patients already in nursinghomes, whose marginal care costs could be zero.

Finally, a patient-based model can accommodatenew information. For future modelling uses, whendata on the duration of the elevated risk of frac-ture become available, the ability to know in which periods the fractures occurred may affectthe results. This can be incorporated into a modelbased on an individual patient but would bedifficult to undertake in a cohort model and still retain accuracy.

In addition, the reviews described in chapters 2and 4 have highlighted several uncertainties relat-ing to costs and quality-of-life changes associatedwith fractures, largely because of the multipleoutcomes that are possible in real life. It is un-certain whether the costs of fractures are depend-ent on the number of previous fractures at a site –for example, whether the cost of treating a secondhip fracture is significantly different from treatingthe first hip fracture. Similarly, the ongoing costsof treating vertebral fractures may differ followinga second vertebral fracture. Indeed, interaction ofall prior fractures in determining the initial andfollow-up treatment costs are not quantified. Inorder that such costs are accurately totalled, a fullpatient history would need to be recorded throughan individual patient-based method.

Similar considerations pertain to the accuracy with which the quality-of-life changes caused byfractures can be calculated when gaps in ourcurrent knowledge are bridged.


Chapter 5

Health economics model


78

Data are required to determine whether thequality-of-life decrements associated with a givenfracture are dependent upon the number ofprevious fractures at that site or elsewhere. Forexample, it may be shown that the decrease inquality of life is different for a first hip fracturethan for a second. Similarly, the loss of quality of life associated with a first vertebral fracture may be different, depending on whether a patienthad previously suffered a hip fracture. If theserelationships are shown to wane with time then the time at which the fractures occurred needs tobe noted. These factors can only be incorporatedinto an individual-based patient model.

The only alternative method by which all data can be taken into consideration is by the use of a decision tree. If a simple model with only fourtransition states is assumed (no fracture event, hip fracture, vertebral fracture and wrist fracture),the tree would require 410 branches over a 10-yearperiod, in order that all conceivable combinationsof events are recorded – resulting in over 1 millionbranches at year 10! This number would be greatlyincreased with the addition of extra states (breastcancer, other fracture states) and would need to be duplicated with the tracking of residential status (community or nursing home). Hence, toreplicate the model using a decision-tree formatwould require over 1 billion branches to maintainthe accuracy of the patient-based approach. This is clearly unmanageable.

Overview of model

In this study, the transition states between whichpatients could move were limited to fracture states(hip, wrist, vertebral, proximal humerus, and deathdue to hip fracture), CHD states (non-fatal andfatal events), breast cancer states (non-fatal andfatal breast cancer) and death from other causes.The probability of a hip fracture causing a patientto reside in a nursing home has been estimated,together with the substantial annual costs thatwould be incurred.

The CHD and breast cancer states were includedbecause there is published literature reporting thattreatments for osteoporosis may also influence theprobability of these diseases. The assumptions thatare used concerning breast cancer risk with HRTand SERMs were given in chapter 2. In the case ofCHD, there is considerable uncertainty over theeffects of HRT. The epidemiological data con-sistently indicate a protective effect but the onlyprospective RCT showed little or no benefit in

women at risk from CHD.153 For the base case, no cardiovascular risk or benefit has been assumed but the assumptions can be changed for sensitivity analysis. Diseases in which there may be possible links with osteoporosis treat-ments, such as Alzheimer’s disease, and venousthrombolic events, that is, cancer, were excludedfrom this study, although the model has beenwritten with spare transition state capacity. Thus,when appropriate, further disease areas may beinvestigated in the future.

The characteristics of the population to beanalysed were flexible. The age, T-score and priorhistory of the population are all user-defined. Here the focus was upon patients with establishedosteoporosis, defined as having suffered a priorfragility fracture; however, the model also has the facility to analyse patients with low bone massbut without prior fracture. In this study, selectedpatient groups were chosen for analysis – forexample, 60-year-old patients with establishedosteoporosis and a T-score of –2.5 SD – althoughthe user may choose to enter whatever patientgroups are desired.

The basic probabilities for moving from transitionstate to transition state have been taken fromepidemiological data, from the UK when possible,and transformed where appropriate. The values ofthese adjustments were in accordance with ratesreported in the published literature.

Having established the transition probabilities, the model simulates the experiences of the cohortunder ‘no treatment’. Outputs are the numbers of life-years gained (discounted at a flexible rate),the numbers of QALYs gained (also discounted at a flexible rate), the discounted costs incurred,and the numbers of each transition state event suffered.

As a patient moves into a transition state, an initial one-off cost is incurred and an ongoing cost that is assumed to last until the end of thesimulation. By using such a methodology, stateswith high ongoing costs can be distinguished from those in which all the costs are incurred in the initial year. In circumstances in which apatient has already suffered the state before, it has been assumed that only the one-off costs will be incurred, with the ongoing costs from thatstate remaining constant. For example, if theconsequences of a vertebral fracture comprised an initial cost of £2000 plus a recurrent cost of£500 annually, a further vertebral fracture in the same individual would cost a further £2000


79

but the recurrent costs would not increase from£500 per year. This may underestimate the costsinvolved but, as mentioned, few data could befound on the additional ongoing costs of second events.

When a patient moves into a transition state, itaffects quality of life. It has been assumed thatthere will be a QALY multiplier effect within thefirst year and a QALY ceiling multiplier that willlast for the remaining years of the simulation. By using this methodology, states from which thepatient will never fully recover can be modelled. It is assumed that, when a patient suffers atransition state for at least the second time, onlythe initial 1-year reduction in quality of life will be taken into consideration. It is noted that, insome cases, this will underestimate the loss inQALYs, for example, second hip or wrist fractureson the opposite side to the first, or a secondvertebral fracture. However, because of lack ofdata, the approach taken was to assume no extraresidual QALY loss from a second incident.

Having established a baseline, ‘no treatment’ cost for the cohort, the incremental effects frompharmaceutical treatments have been calculated.

The duration and the acquisition cost of eachtreatment are user-definable. The efficacy of eachtreatment is modelled by the use of RRs on enter-ing a transition state. It is expected that a cohortusing a treatment with an RR of 0.5 for hip frac-ture would, in the next period, have half thenumber of hip fractures as the same cohortreceiving ‘no treatment’ (RR = 1), assuming an equal death rate.

The RRs have been subjected to meta-analysis foreach treatment whenever possible, using publishedRCTs with the number of fractures as an endpoint(chapter 2). The effectiveness of each treatmenthas relatively large uncertainties but the meta-analyses have provided distributions and 95% CIs.

In addition to the RRs for efficacy of treatment,the model incorporates offset times, defined as the time from when treatment is stopped to when the RR returns to unity compared with ‘notreatment’. It is assumed that the RR returns tounity in a linear manner during the offset time.The incorporation of offset times is crucial inaccurately modelling those treatments that areconsidered to have long residual effects.

Each treatment option has also been assigned costsadditional to drug acquisition, namely GP visits,

assumed to be two per annum, and BMD scans,assumed to occur in years 2 and 5 of treatment. In sensitivity analyses, when treatment was givenfor 10 years, the second BMD scan was assumed to occur at year 10. Lack of compliance was alsomodelled, assuming that the patient incurs 3 months of drug costs but receives no healthbenefits. It has been assumed that, for the year in which death occurred, the QALYs gained arehalf those for the prior year, the costs incurred are equal to half of the ongoing annual costs, and only 50% of the drug acquisition cost is paid.

A complex methodology for estimating the cost-effectiveness of each drug was employed, in orderfor a distinction to be made between variations inthe results due to random events (e.g. prematuredeath) and those caused by the uncertainties inthe true RRs for the efficacy of each drug, asindicated by the 95% CIs.

The basic design of the Sheffield Health Eco-nomics Model for Osteoporosis (SHEMO) issimilar, in many ways, to the conventional Markovmodels used in osteoporosis, in which patients pass through states using a set of transition prob-abilities, and each state has its associated costs,mortality rates and health state utility values.However, SHEMO differs in two crucial respects to the conventional cohort Markov design. First,individual cases pass through the model one at a time and then a mean estimate is taken ofcosts, mortality and QALYs for each cohort. Theadvantage of this individual level approach is thatit is able to take account of a patient’s history interms of factors, such as prior fracture (a key factor in fracture rates) and residential status, that can be used to determine the likelihood of an event occurring in the next period. Also, by recording the number of events, and the period in which they occur, the model can provideadditional output, once new data become available,on factors such as the interaction of fractures interms of costs and QALYs, and for how long theincreased risk of subsequent fractures persists.

SHEMO also differs from cohort Markov models in that it is a stochastic model that formallyincorporates the uncertainties that underlie keyparameters. The distribution of values for clinicalefficacy is used to reflect current uncertainty, based on the 95% CIs estimated from a systematicreview of the published literature. The modelworks by undertaking the individual level simu-lations in cohorts and, for each cohort, the RRs are re-sampled using a Monte Carlo simulation.



80

The key features are:

• an individual patient approach• accommodation of multiple states: for example,

prior fracture, residential status, change in status• account taken of uncertainty: for example,

in efficacy.

The process of estimation is in two parts. The first estimates the relationship between the inputs of the model and the outputs, in terms of costs and QALYs. To do this, approximately 80 different combinations of RR values for each clinical condition (the fractures, CHD and breast cancer) were selected for each agegroup. For each combination, 8000 patients were simulated to give mean costs and QALYestimates. By undertaking runs that simulated 8000 patients, it was possible to remove a largeproportion of ‘noise’. The relationship betweenthese model inputs and costs and QALYs wasestimated using a non-parametric technique called a Gaussian process.339 This effectivelyproduces a formula that allows instant calcu-lation of the expected QALY and costs for any parameter set.

The second phase of the estimation processexamines the consequences of the uncertaintyaround the efficacy estimates for each treatment.For each treatment, 1000 values of efficacy for eachtype of fracture (plus CHD and breast cancer forsome treatments) were selected using Monte Carlomethods. From the 1000 samples of parameterpoints, the model formulated in the first phase was used to generate estimates of 1000 costs andQALYs. These formed the basis of the estimatedmean cost per QALY, and the associated CIs,compared with ‘no treatment’.

The mean cost per QALY was calculated as the mean cost difference divided by the meanQALY difference for the 1000 points and for ‘notreatment’. The CIs were calculated by ranking thecost per QALY from each of the 1000 parametersand ascertaining the 90% and 80% CIs. These CIs reflect the genuine uncertainty around theestimate rather than random noise. As the resultshave been generated from a formula, any differ-ences in the mean cost per QALY and the CIsbetween treatments are due solely to the RRsaround efficacy.

The advantage of using the Gaussian processtechnique is that, given the same startingassumptions, the results for a new drug withdefined RRs can be instantly calculated.

Population of the modelPopulation start ageThe model had the flexibility to allow the age ofthe cohort of patients to be set at yearly intervalsbetween 45 and 109 years of age. For our purposes,women were chosen at ages 50, 60, 70 or 80 yearsof age.

Osteoporotic fractureThe present study considers fractures of the spine, hip, proximal humerus and distal forearm.Established osteoporosis is defined as an individualwith one or more of these fractures, having a T-score at the femoral neck that is below the diagnostic threshold.

Distributions of fracturesThe starting distribution between states for estab-lished osteoporosis was taken from the incidenceof fracture presented in chapter 4 and summarisedin Table 56. For each year over age 50 years, theexpected cumulative number of fractures per sitewas calculated.

These were then proportioned to provide thepercentages shown in Table 57. For example, 8% of osteoporotic fractures up to the age of 50 yearswere hip fractures. This figure rose with age, andhip fractures accounted for 21% of all osteoporoticfractures at age 80 years. Thus, in each cohort of 100 individual patients at age 70, 11% wereassumed to have had hip fractures, 19% vertebralfractures, 56% wrist fractures and 14% proximalhumerus fractures.

This approach is likely to produce some bias,caused by patients with more than one priorosteoporotic fracture. For example, in an extreme case in which all 80-year-old patients had one prior hip, vertebral, wrist and proximalhumerus fracture, the starting distribution would be set with 25% for each fracture, despite100% of people having sustained a hip fracture.The alternative strategy would be to computeprobabilities of first and subsequent fractures –data that are not available for the UK. Asmentioned previously (in chapter 2), suchprobabilities would need to be adjusted for secular trends in mortality.

Initial BMD-score of the populationThe initial BMD in terms of a T-score, (number of SDs below the BMD of a young female adult) can be user defined. Osteoporosis is defined as a T-score of –2.5 SD unless otherwise indicated.


81

Discount ratesThe discount rate for costs was set at 6% perannum, in accordance with published guidelines.340

The default discount rate for QALYs was set at1.5% per annum.341 However, because of debateabout the value that should be used, a sensitivityanalysis, run in parallel, used a discount value of6% per annum.

Default state transitionprobabilitiesIn this study, the following transition states wereused in the model.

1. Osteoporotic (never previously been in anyother state)

2. Sustained a hip fracture3. Hip fracture and confined to nursing home4. Death due to a hip fracture5. Sustained a vertebral fracture6. Sustained a wrist fracture7. Sustained a proximal humerus fracture8. Contracted breast cancer9. Death due to breast cancer10. Sustained a non-fatal CHD incident11. Death due to a CHD incident12. Death due to other causes

There was also a ‘no event’ state, which signifiedthat a patient did not have an event which wouldbe associated with a change of state.

The model could accommodate 25 different states, so that conditions that were suspected but currently unproven to have RRs associated with osteoporosis treatment could be entered into a future model if new evidence was obtained.

The model simulated each patient from entry into the model until death, age 110 or at amaximum period specified by the user. In thisstudy, the model used a time frame of 10 years, and 15 years in the sensitivity analyses.

Each state was reviewed in chapter 4 with details of the assumed probabilities of moving into that state. The probability of ‘no event’ is one minus the sum of probabilities for moving to all states. The states are sum-marised briefly below.

OsteoporoticThis state is reserved for those who have notsuffered one of the remaining defined states.Hence the probability of moving into this statefrom any other state is zero.


TABLE 56 Annual fracture risk by age at the sites shown

Age range (years) Annual fracture risk (%)

Hip Spine Wrist Proximal humerus

50–54 0.041 0.160 0.255 0.058

55–59 0.050 0.087 0.374 0.085

60–65 0.083 0.139 0.467 0.136

65–69 0.157 0.178 0.573 0.126

70–74 0.485 0.461 0.699 0.246

75–79 0.707 0.467 0.697 0.306

80–84 1.437 0.503 0.749 0.372

85–89 2.761 0.884 1.001 0.362

90+ 3.851 1.232 0.919 0.391

TABLE 57 The estimated starting distributions of established osteoporosis at various ages

Fracture site Estimated starting distributions of osteroporosis (%) at ages (years)

50 60 70 80

Hip 8 8 11 21

Vertebral 31 22 19 22

Wrist 50 57 56 43

Proximal humerus 11 13 14 14


82

For patients in this state, the probability is zero,with ‘no event’ signifying that a patient remainshealthy although osteoporotic. As the focus of thisstudy is on patients with established osteoporosis,this state was not populated in this model.

Fracture risksThe estimated figures for the average populationare shown in Table 56. The average population riskswere adjusted for a population with osteoporosis.The risk of fracture in the general population wasadjusted from the known relationship betweenBMD and fracture risk, assuming that BMD wasmeasured at the femoral neck. The gradients ofrisk/SD decrease in BMD were taken from a meta-analysis16 that was similar to those derived from theSOF study.9 The predicted number of wrist andvertebral fractures was slightly lower than thosepredicted by use of the SOF figures.

The fracture risks were computed for individualswith a T-score of –2.5 using these gradients and thepattern of change of BMD with age described inchapter 4.

Death due to hip fractureIt was assumed that 48% of all deaths in the first year associated with hip fracture are causallyrelated to the fracture and would therefore beavoided by preventing hip fracture (chapter 4).The attributable fraction was changed in asensitivity analysis for the reasons described in chapter 4.

First entry to nursing home after hip fractureProbabilities taken from the second Anglian audit of hip fracture were used, as detailed in chapter 4.

Death due to vertebral fractureNo excess mortality was assumed other than thataccounted for by low BMD. It was assumed thatinterventions that increase the BMD of a patientwill not change the T-score (BMD)-adjusted risk of death from other causes.

Developed breast cancerThe risks of breast cancer, taken from the cancerregistrations in England and Wales, were adjusteddownwards to accommodate the evidence that lowBMD is associated with a lower breast cancer risk.

Death due to breast cancerRisks of death for the UK were used. It wasassumed that the risk of death following breastcancer in women with osteoporosis was similar to that in the general population.

Death due to CHDRisks of death for the UK were used. It wasassumed that the risk of death from CHD inwomen with osteoporosis is similar to that in the general population.

Sustained a non-fatal CHD eventData were derived from those for England. It was assumed that the probability of CHD inpatients with osteoporosis is similar to that in the general population.

Death due to other causesThese were computed from interim life tables andadjusted for deaths due to CHD and breast cancer.It should be noted that excess mortality wasassumed for low values of BMD.

Adjustments to the defaulttransition probabilitiesThe model had the facility to allow prior patientstates to influence the transition matrix. This wasneeded since the risk of a secondary fracture ishigher than the risk of an initial fracture.

A summary of each state is given below, togetherwith the transition probabilities that can be altered.

OsteoporoticThis state does not impact upon any transitionprobabilities.

All fracture statesA prior fracture substantially increases the risk of subsequent fractures. The meta-analysis fromKlotzbeucher and colleagues has been used withsome additional assumptions.263 It was assumedthat future fractures at the proximal humerus were equivalent to future fractures that were in the non-spinal category. It was also assumed thatthe proximal humerus had the predictive powerequal to that of the ‘other’ category. All popu-lations were assumed to be peri/postmenopausal.There were no prior studies on the future effectthat hip fractures might have on wrist fractures. As a conservative estimate, this risk was set at 1.4, equivalent to the lowest RR of all otherfracture sites.

It was assumed that, for individuals who sufferedfractures at two different sites, only the greatestrisk adjustment would be applied. For example,were a patient to have both a prior hip and wristfracture, the RR adjustment for a vertebral fracturewould be 2.5 (from the hip fracture) and for a


83

second wrist fracture would be 3.3 (from the wrist fracture).

Contracted breast cancerThe model has the facility to increase the risk ofcontracting breast cancer following an earlier non-fatal breast cancer incident. Because of a paucity ofprospective data, this was set to RR = 1 in the basecase but altered for the sensitivity analyses. Achange in risk was assumed to revert to 1.0 whentreatment was stopped (i.e. offset time = 0).

Suffered a CHD incidentThe model has the facility to increase the risk of a CHD incident following an earlier non-fatalCHD incident. Because of lack of data, this was set to RR = 1 in the base case but altered forsensitivity analyses. A change in risk was assumed to revert to 1.0 as soon as treatment stopped.

ComplianceIt was assumed, in consultation with clinicians, that the patient, if non-compliant, would incur 3-months’ drug intervention costs but accrue no health benefit.

Treatment

For each therapeutic intervention, the efficacy was assumed to equal the estimate of the entirefrequency distribution of RR derived by meta-analysis (chapter 2). The effect of treatment onfracture probability was instantaneous and per-sisted unchanged throughout the treatmentperiod. It was assumed therefore that the effective-ness did not change with time. There is increasingevidence that anti-fracture efficacy is greater in thefirst year of treatment than thereafter (see chapter2) and thus, as the duration of treatment increases,the assumption of a consistent risk reductionbecomes unsafe. This was one of the reasons for selecting a 5-year treatment time that corres-ponded to the duration of exposure in RCTs,particularly those undertaken in the past 10 years.

The treatment effect was not bounded by the 95% CI but the entire distribution of effect wasincluded in the analysis. In other words, efficacywas assumed to vary in individuals drawn from thecohort according to the probability density. Formany treatments, the 95% CI for efficacy exceededa value of 1. Since osteoporosis is a systemic diseaseand the risk of any fragility fracture at the spine,wrist, forearm or shoulder is increased in thepresence of a prior fracture at any of these sites,the notion that hip fracture rates may be increased

when RCT evidence suggests that other fragilityfractures are significantly decreased is counter-intuitive. For the reasons discussed earlier, a major reason for the paucity of robust infor-mation on hip fracture risk is related to theexpense of undertaking such studies and theregulatory framework, which does not encouragesuch studies for registration. For this reason,sensitivity analyses were undertaken in two ways relating to the effectiveness of treatment.When there was a small effect or no effect on hip fracture risk, the RR was set at an absolutevalue of 1.0. When there was a substantial effect on hip fracture risk but the CIs exceeded 1.0, the estimate was set at the point estimate without CIs.

When treatment was stopped, the effect oftreatment was assumed to wane in a linear mannerover time. The persistence of some therapeuticeffect is well documented with some interventions(see chapter 2). The offset time was assumed to be 5 years for all interventions except calcium andcalcitonin. For these agents, a 3-year offset timewas assumed. In other words, the fracture riskincreased progressively after stopping treatmentand, at the end of the offset period, was the sameas that predicted in untreated individuals. Offsettime was changed in the sensitivity analyses.

In the base case no interaction was assumedbetween treatment and non-skeletal outcomes. As mentioned previously, patients with osteo-porosis were assumed to have a lower risk of breast cancer. Such data were derived from epidemi-ological estimates and may therefore be subject to bias. When treatments had an effect on breastcancer risk, this assumption was tested by sensi-tivity analysis. In the case of HRT, epidemiologicalevidence also indicated that it might afford sub-stantial cardiovascular protection (see chapter 2).Again, this was derived from epidemiological dataso was not considered in the base case. Base-caseassumptions and sensitivity analyses for efficacy are shown in Tables 58 and 59.

All of the agents under consideration have beenshown to have side-effects. In most instances, theprevalence was not well documented, nor were theconsequences known for quality of life expressedin utilities (chapter 2). Also, the impact of side-effects on compliance is conjectural. Adverseeffects have not been included in the analysis,although it should be recognised that even smallgains or decrements in quality of life caused byside-effects could have a marked impact on cost-effectiveness.



84

TABLE 58 Efficacy of agents on fracture risk at the sites shown for the base case (A) and sensitivity analyses (B/C)

Agent RR (0.95% CI)

Spine Hip Forearm Humerus

Alendronate A 0.544 (0.448 to 0.659) 0.611 (0.392 to 0.951) 0.866 (0.672 to 1.115) 0.825 (0.736 to 0.926)

Bisphos- A 0.526 (0.445 to 0.637) 0.672 (0.459 to 0.983) 0.833 (0.659 to 1.054) 0.824 (0.745 to 0.913)phonatesa Bb 0.575 (0.490 to 0.675) 0.620 (0.368 to 1.042) 0.566 (0.377 to 0.848) 0.813 (0.693 to 0.954)

Alfacalcidol A 0.459 (0.149 to 1.414) 0.249 (0.009 to 6.768) 1.042 (–c) 0.193 (0.007 to 5.068)B 0.459 (0.149 to 1.414) 1.0 1.0 1.0

Calcitonin A 0.308 (0.113 to 0.838) 0.681 (0.145 to 3.198) 0.947 (0.197 to 4.565) 0.553 (0.224 to 1.254)B 0.308 (0.113 to 0.838) 1.0 1.0 0.553C 0.308 (0.113 to 0.838) 0.63 (0.440 to 0.900d) 1.0 0.553

Fluoride A 0.350 (0.253 to 0.486) 1.778 (0.758 to 4.170) 0.766 (0.371 to 1.583) 0.975 (0.782 to 1.217)B 0.350 (0.253 to 0.486) 1.0 1.0 1.0

Calcium A 0.550 (0.330 to 0.930) 1.0 1.0 1.0Be 0.550 (0.330 to 0.930) 0.738 (0.600 to 0.908) 0.739 (0.687 to 0.917) 0.739 (0.687 to 0.917)

a Aggregate of bisphosphonate datab Patients with prior fracturec Confidence estimate not modelled due to high range of risk (0.021 to 51.993)d Observational estimatee Calcium plus vitamin D

TABLE 59 Efficacy (RRs and 95% CIs) of raloxifene and HRT on fracture risk, CHD and breast cancer showing the base case andsensitivity analyses

Base case (A) RR (95% CI) of fracture RR (95% CI) RR (95% CI) ofor sensitivity

Spine Hip Forearm Humerusof CHDa breast cancera

analysis(B, C, D, E)

RaloxifeneA 0.596 1.141 0.887 0.920 1.00 0.35

(0.516 to 0.688) (0.663 to 1.966) (0.684 to 1.151) (0.792 to 1.068) (0.21 to 0.58)

B 0.596 1.00 1.00 1.00 1.00 0.35(0.516 to 0.688) (0.21 to 0.58)

C 0.596 1.00 1.00 1.00 0.80b 0.35(0.516 to 0.688) (0.21 to 0.58)

OestrogenA 0.583 1.00 1.00 1.00 1.00 1.00

(0.262 to 1.301)

B 0.583 0.86c 0.32c 0.63c 1.00 1.00(0.262 to 1.301) (0.42 to 1.75) (0.13 to 0.78) (0.45 to 0.89)

C 0.583 0.86c 0.32c 0.63c 1.00 1.00(0.13 to 0.78) (0.45 to 0.89)

D 0.583 0.86c 0.32c 0.63c 0.66c 1.00(0.13 to 0.78) (0.45 to 0.89) (0.53 to 0.84)

E 0.583 0.86c 0.32c 0.63c 0.66c 1.35c

(0.262 to 1.301) (0.42 to 1.75) (0.13 to 0.78) (0.45 to 0.89) (0.53 to 0.84) (1.21 to 1.49)

a Offset time = 0b Computed from lipid changesc Observational estimate


85

The failure to include beneficial effects onsymptoms is most problematic in the case of HRT,since HRT has been proven to reduce symptomsassociated with the menopause. Indeed, healtheconomics analyses suggest that treatment for this indication alone is cost-effective.2 It wasdecided not to include such effects for severalreasons. First, the use of HRT for menopausalsymptoms is given for months rather that theseveral years required for the management ofosteoporosis. Second, a very small minority of patients at menopause are expected to have osteoporosis. Third, the assumptionsconcerning efficacy were based on epidemi-ological estimates with their own uncertainties, so that the inclusion of an effect on menopausalsymptoms provided an unnecessary level of

complexity for uncertain gains. The additionalpotential extra-skeletal benefits of HRT (andSERMs) on cardiovascular disease wereaccommodated in a sensitivity analysis.

All analyses are based on the 10-year time framerather than over a lifetime, unless otherwise stated.In the context of treatments that are presentlydeveloped for 3–5 years, a 10-year interval was con-sidered to be more appropriate. It takes account ofthe intervention period as well as the offset time oftherapeutic effect once treatment is stopped.44 Inaddition, the predictive value of risk factors such as low BMD becomes less over intervals greaterthan 10 years.42 A time frame of 15 years was used in sensitivity analysis to model a change in the assumption relating to offset time.



87


Analytical approachThe results for each drug at each age group arepresented in terms of a central estimate of cost per QALY gained and a cumulative frequencydistribution represented in the tables by the 90% CIs. Note that the CI is not that of the central estimate of cost-effectiveness but of therange of costs per QALY gained that is incurred by 90% of runs sampled over the range of efficacyfor the intervention. An example is provided inFigure 19 that shows the cost-effectiveness ratio of a hypothetical agent in women aged 60 years.The mid-point estimate of cost-effectiveness was£22,557. The cost-effectiveness ratio varied be-tween £13,562 and £546,604. In 90% of theestimates, the cost-effectiveness ratio lay between £17,362 and £40,779.

In cases where the cost-effectiveness curveintersects with the y-axis, the intercept denotes the proportion of estimates where treatment isdominant (i.e. ‘cost-savings with health benefits’

compared with ‘no treatment’). Where the curve does not reach 100%, the value for the y-axis denotes the proportion of estimates that are dominating (i.e. ‘increasing costs’ and‘detrimental to health’).

Because of the very wide CIs generated by analysis of trials of calcitriol (RR = 0.87; 95% CI, 0.00 to 218.4 for hip fracture), and becausethere was no effect on vertebral fracture risk, this intervention was omitted from the analyses.However, alfacalcidol has been included, althoughthis agent is not a licensed treatment for osteo-porosis in the UK.

Costs for different interventions are summarised in Table 60.

The results from the base-case assumptions aredescribed below. These discount costs at 6% andQALYs at 1.5% for each treatment. The total costsand QALYs are given for 100 patients and can be compared with those for untreated patients

Chapter 6

Results

10 20 30 40 50 60 70 80

Cost/QALY gained (£000)

100

90

80

70

60

50

40

30

20

10

0

Cumulative frequency (%)

FIGURE 19 Distribution of cost-effectiveness of a hypothetical agent in women aged 60 years with established osteoporosis. Horizontallines denote the cost-effectiveness ratio of 5%, 50% and 95% of the cohort (£17,362, £22,557 and £40,779, respectively)

Results

88

presented in Table 61 to derive the marginal costsand QALYs. Each treatment was given to a popu-lation of women with osteoporosis, that is, with apredetermined ratio of individuals with priorfractures of different types. Results of cohorts withspecific fracture types are shown subsequently insensitivity analyses and as clinical vignettes.

In this study, a threshold of £30,000/QALY gained for cost-effectiveness has been used:

that is, treatments that have a cost–utility ratio of £30,000 or less are considered cost-effective.342

Since the methodology gives ‘CIs’ for cost-effectiveness, this permits several categories to be derived, based on the 90% interval, as shown in Table 62. Note that the 90% CI describes thecost-effectiveness ratio computed in 90% of the samples and not the confidence estimate of the mid-point estimate. A grade of A or B was considered to be cost-effective for the purposes of this study.

Specific treatments

RaloxifeneRaloxifene significantly reduced the risk ofvertebral fracture (RR = 0.60) but had nosignificant effect on appendicular fractures. Forthe base case, a significant effect on breast cancerwas assumed (RR = 0.35), in line with RCTevidence and, in the absence of RCT data, aneutral effect on CHD.

It was not cost-effective to treat women withestablished osteoporosis. At age 50 years, the cost per QALY gained was £572,125 (Table 63). The cost per QALY decreased at age 60 years butthereafter increased with age. This is because theassumed effectiveness on hip fracture is negative(RR = 1.14). The adverse effect of hip fracture risk with age was not compensated by additionaldividends from breast cancer prevention, at least over the 10-year time frame modelled.

When the assumption was made that the effects of raloxifene on appendicular fractureswere neutral (i.e. RR set to 1.0), then cost-effectiveness improved, and improved withadvancing age except at age 80 years (Table 64).Cost-effectiveness remained on average above the threshold of £30,000/QALY gained in 90% of the treated cohort at all ages.

TABLE 60 Costs of different interventions

Intervention Annual cost (£)

No treatment 0

Calcium 40

Fluoride 48

Calcium plus vitamin D 55

HRT 58

Alfacalcidol 157

Etidronate 163

Raloxifene 257

Alendronate 334

Calcitonin 2314

TABLE 61 Costs and QALYs for ‘no intervention’ for a cohort of100 patients over 10 years

Age (years) Total cost Total QALYs(£000)a (000)b

50 92.0 702.5

60 133.1 669.4

70 285.3 530.1

80 645.6 366.7

a Discounted at 6%b Discounted at 1.5%

TABLE 62 Classification of cost-effectiveness

Grade Description Cost-effectiveness (£/QALY gained) 90% CI

Mid-point Lower Upper

A* Always cost-effective < 0 < 0 < 0

A Always cost-effective < 30,000 < 30,000 < 30,000

B Probably cost-effective < 30,000 < 30,000 > 30,000

C Possibly cost-effective > 30,000 < 30,000 > 30,000

D Never cost-effective > 30,000 > 30,000 > 30,000


89

Since RCTs showed that raloxifene had significanteffects on surrogate markers of CHD, a 20% RRreduction in cardiovascular disease was examined.This resulted in a marked improvement in cost-effectiveness (see Table 64) that improved with age. At age 70 years or more, raloxifene was cost-effective in nearly 90% of treated women. The cost-effectiveness acceptability curve for thisscenario is shown in Figure 20.

Cost-effectiveness ratios improved when patients at higher risk were treated (T-score < 2.5 SD; seepage 100).

A focus of the brief for raloxifene was to examineits cost-effectiveness for prevention as well as fortreatment of established osteoporosis. Raloxifenewas not found to be cost-effective except in thevery elderly, so that prevention would be even lesscost-effective, since the risks of fracture are lowerin patients without prior fractures.

HRTThe base case for HRT assumed efficacy only on vertebral fracture risk (RR = 0.58) and, in


TABLE 63 Cost-effectiveness of using of raloxifene in women with established osteoporosis according to age: base case

Age (years) Cost Total Cost/QALY gained (£000) Grade(£000)a QALYsb

Midpoint 90% CI

50 235.1 702.8 572.1 61.6 dominated D

60 280.2 670.9 94.7 29.6 dominated D

70 441.1 530.9 187.0 11.6 dominated D

80 815.0 366.0 dominated dominating dominated D

a Discounted at 6%; b discounted at 1.5%

TABLE 64 Cost-effectiveness of using of raloxifene in women with established osteoporosis according to age (raloxifene is assumed notto affect the risk of appendicular fracture)


Midpoint 90% CI

Neutral effect on appendicular fracture50 232.2 703.5 148.5 126.8 180.8 D60 273.0 672.3 47.8 35.2 71.0 D70 411.8 532.8 46.4 31.4 59.0 D80 754.0 368.2 76.1 50.0 223.0 D

Effect on cardiovascular disease50 232.2 703.5 148.1 126.8 180.8 D60 273.1 672.4 46.6 31.9 70.1 D70 414.7 535.1 25.6 20.2 32.0 B80 754.1 373.9 15.2 11.2 20.6 A

a Discounted at 6%; b discounted at 1.5%

100

80

60

40

20

00 10 20 30 40



FIGURE 20 Distribution of cost-effectiveness of raloxifene inwomen aged 70 and 80 years with established osteoporosis (●●, 70 years; ●, 80 years)

Results

90

line with RCT data, no effect on appendicularfractures or CHD. Note that the 95% CI forvertebral fracture efficacy crossed unity. Underthese assumptions, it was not cost-effective inwomen aged 50 years (Table 65). At age 60 years or more, the mid-point estimate for cost per QALY gained was less than £30,000.

The assumptions on effectiveness were varied in sensitivity analyses. The first analysis was toassume that HRT decreased the risk of appen-dicular fractures in line with the extensive epidemiological data. Note that the RR of hip fracture was reduced but the 95% CI crossed unity (RR = 0.86; 95% CI, 0.42 to 1.75).Under this assumption, cost-effectiveness improved (Table 66a). On average, cost-effectiveness scenarios were found at all ages. Cost-effectiveness improved still further when point estimates of efficacy were used to model hip fracture outcomes (Table 66b), that is, for this purpose, the RR was pegged at 0.86 withoutthe use of confidence estimates. In this way, therisk of hip fracture could not increase over the control risk. The improvement in cost-effectiveness was more marked with advancing age and the grading of cost-effectiveness improvedfrom B to A at age 70 years and from B to A*

at age 80 years.

In the absence of RCT data, the model aboveassumed neutral effects on CHD and breast cancer. Population of the model with effects ofHRT on CHD, as observed in epidemiologicalstudies, improved cost-effectiveness (Table 66c).Little improvement was observed at age 50 yearsbecause of the low absolute risk of cardiovasculardisease. At age 60 years or more, treatment wasalways cost-effective.

The addition of a significant increase in the risk of breast cancer (RR = 1.35) had a modest

impact on the central estimate of cost-effectiveness (Table 66d). Overall, the principal effect of adding cardiovascular plus breast cancer risk was to improve the cost-effectiveness ratiomarginally: that is, the benefits of cardiovascularprotection largely offset the risks of breast cancer.In the cost-effectiveness curves shown in Figure 21,it is assumed that HRT has both cardiovascularbenefits and breast cancer risks. Note that cost-effectiveness was not observed in all patients.

CalciumIn the base case for calcium, only effects on verte-bral fracture were considered, in accordance withthe methodology chosen for the meta-analysis ofRCTs (RR = 0.55). It was assumed, therefore, thatcalcium had no effect on appendicular fractures.An offset time of 3 years was modelled. On aver-age, calcium was cost-effective from age 60 years(Table 67a). Cost-effectiveness acceptability curves are shown for the cost-effective scenario in Figure 22.

The evidence from RCTs indicated, however, that calcium given with vitamin D decreased the risk of appendicular fractures in women. In these studies, BMD was not systematicallyassessed and was therefore omitted from the meta-analysis and base-case analysis. When these effects were included in the model(using a price for calcium and vitamin D), it was, on average, cost-effective to use calcium with vitamin D at all ages. Cost-effectiveness was below the threshold of £30,000/ QALY gained in > 90% of patients aged 60 years or more (Table 67b). Cost-savings occurred at age 80 years.

Cost-effectiveness acceptability curves are shownfor all ages in Figure 23. Cost savings occurred in43% of individuals at age 70 years and in 98% atage 80 years.

TABLE 65 Cost-effectiveness of using HRT in women with established osteoporosis according to age


Midpoint 90% CI

50 144.5 703.4 61.9 30.9 dominated D

60 186.7 671.3 27.8 17.2 dominated B

70 330.6 532.0 24.1 13.3 dominated B

80 681.8 368.6 19.9 6.1 dominated B



91


TABLE 66 Sensitivity analysis of cost-effectiveness of using HRT in women with established osteoporosis according to age


Midpoint 90% CI

(a) HRT affects appendicular fractures50 140.4 704.2 29.3 10.7 dominated B

60 176.7 672.8 12.7 3.6 dominated B

70 313.5 534.3 6.7 dominating dominated B

80 633.1 373.2 dominating dominating dominated B

(b) HRT has fixed effect on appendicular fracture50 139.3 704.4 25.1 16.8 65.1 B

60 174.0 673.1 10.9 7.7 20.7 A

70 304.7 534.4 4.5 2.5 12.0 A

80 611.7 373.7 dominating dominating dominating A*

(c) Additional benefits in CHD included50 139.3 704.4 25.1 16.8 65.1 B

60 174.1 673.3 10.6 7.6 20.5 A

70 304.8 536.4 3.1 1.8 7.9 A


(d) Additional risk of breast cancer included50 140.4 704.2 29.4 10.7 dominated B

60 176.7 672.9 12.4 3.6 dominated B


80 633.1 376.4 dominating dominating dominated B


100

80

60

40

20

00 20 40 60 80 100



FIGURE 21 Distribution of cost-effectiveness of HRT in women aged 50–80 years with established osteoporosis (■■, 50 years;■, 60 years; ●●, 70 years; ●, 80 years)

Results

92

CalcitoninIt was not cost-effective to use calcitonin for long-term treatment of established osteoporosis,according to the data derived from the meta-analysis (Table 68a). An offset time of 3 years wasused in the model. The fact that the assumedeffects on hip fracture (RR = 0.68) ranged from0.15 to 3.20 is relevant. When neutral effects onhip fracture were assumed, only minor effects were noted and cost-effectiveness was still not seen at any age (Table 68b). As reviewed earlier in chapter 2, the mid-point estimate for the effects of calcitonin on hip fracture risk from RCTs is similar to that from the epidemiologicaldata, that is, 0.63. The confidence estimate for the case–control analysis indicates a significantassociation between exposure to calcitonin and hip fracture risk (95% CI, 0.44 to 0.90). Theinclusion of this effect in the model improved cost-effectiveness but not sufficiently to impact on the grade of cost-effectiveness (Table 68c).

AlendronateThe most extensive RCT information base exists for alendronate. The agent has been shownto significantly reduce hip fracture (RR = 0.61),vertebral fracture (RR = 0.54) and humeral frac-ture (RR = 0.83). The effect on forearm fractures(RR = 0.87) was not significant. The cost-effectiveness ratio improved with age. Treatmentwith alendronate became cost-effective at age 70 years or more. At age 80 years, treatment wasalways cost-effective (Table 69 and Figure 24).

BisphosphonateThere was little evidence of differences in efficacy between the bisphosphonates (see

chapter 2), and for this reason, data were pooled.Mid-point estimates were comparable to those of alendronate for spine and humeral fractures but slightly less favourable for hip fracture andslightly more favourable for forearm fracture (see Table 60). With a treatment cost of £163 yearly (the price of etidronate), treatment was not cost-effective at age 50 years but became

TABLE 67 Cost-effectiveness of using calcium or calcium with vitamin D in women with established osteoporosis according to age


Midpoint 90% CI

(a) Calcium: effects on vertebral fractures only50 136.1 703.5 44.0 28.5 121.4 C

60 178.5 671.5 21.4 15.0 44.7 B

70 322.0 532.4 15.9 10.8 41.5 B

80 671.5 368.3 16.3 5.1 124.3 B

(b) Calcium + vitamin D: effects on appendicular fractures also50 136.7 704.8 19.8 14.2 33.6 B

60 169.9 673.4 9.2 6.0 16.1 A

70 287.9 535.6 0.4 dominating 6.7 A



100

80

60

40

20

00 20 40 60 80



FIGURE 22 Distribution of cost-effectiveness of calcium in women aged 60–80 years with established osteoporosis (■, 60 years; ●●, 70 years; ●, 80 years)


93


100

80

60

40

20

00 10 20 30 40 50 60



FIGURE 23 Distribution of cost-effectiveness of calcium plus vitamin D in women aged 50–80 years with established osteoporosis (■■, 50 years; ■, 60 years; ●●, 70 years; ●, 80 years)

TABLE 68 Cost-effectiveness of using calcitonin in women with established osteoporosis according to age


Midpoint 90% CI

(a) Results of meta-analysis50 1138 704.1 641 205 dominated D

60 1161 672.6 301 118 dominated D

70 1256 535.9 168 55 dominated D

80 1472 373.4 124 27 dominated C

(b) Assuming neutral effect on hip fracture50 1138 704.1 685 512 1413 D

60 1162 675.6 323 248 551 D

70 1263 534.2 235 159 777 D

80 1490 370.2 245 142 533 D

(c) Effects on hip fracture from epidemiological data50 1132 705.6 341 247 1450 D

60 1147 675.0 182 175 1207 D

70 1210 538.6 108 78 190 D

80 1363 377.0 70 44 146 D


Results

94

TABLE 69 Cost-effectiveness of using alendronate in women with established osteoporosis according to age


Midpoint 90% CI

50 258.2 705.2 61.4 45.5 111.6 D

60 287.2 673.8 35.3 24.4 63.7 C

70 387.0 537.8 13.3 6.2 37.0 B

80 633.4 376.5 dominating dominating 21.3 A


100

80

60

40

20

00 10 20 30 40 50 60 70



FIGURE 24 Distribution of cost-effectiveness of alendronate in women aged 70 and 80 years with established osteoporosis (●●, 70 years; ●, 80 years)

TABLE 70 Cost-effectiveness of using bisphosphonate in women with established osteoporosis


Midpoint 90% CI

Intervention costs £163/year50 183.5 705.1 36.1 26.0 66.1 C

60 215.2 673.5 19.8 13.0 36.4 B

70 324.5 536.9 5.7 0.9 21.1 A


Intervention costs £334/year50 259.0 705.1 65.8 48.4 117.8 D

60 289.5 673.5 37.8 26.0 66.5 C

70 395.3 536.9 16.1 7.9 40.1 B

80 653.7 375.4 0.9 dominating 28.2 A



95

cost-effective at age 60 years and above. Cost-savings occurred from age 80 years (Table 70). The cost-effectiveness acceptability curves for the cost-effective scenario are shown in Figure 25.

With a higher cost of treatment (£334 annually,equivalent to alendronate), cost-effectiveness wasseen at age 70 years and above (Table 70).

An analysis using risk estimates confined topatients with established osteoporosis had slightly

lower cost-effectiveness ratios but did not affect the grading (Table 71).

FluorideFluoride significantly decreased the risk ofvertebral fracture (RR = 0.35) but appeared toincrease the risk of hip fracture (RR = 1.78),although this was not statistically significant.Fluoride was not cost-effective at any age using the RCT data (Table 72). When it was assumed that the effect of fluoride on hip and otherappendicular fractures was neutral (RR = 1.0),


100

80

60

40

20

00 10 20 30 40 50



FIGURE 25 Distribution of cost-effectiveness of bisphosphonate in women aged 60, 70, and 80 years with established osteoporosis (■, 60 years; ●●, 70 years; ●, 80 years)

TABLE 71 Cost-effectiveness of using bisphosphonate in women with established osteoporosis based on meta-analysis of women onlywith prior fracture


Midpoint 90% CI

Intervention costs £163/year50 182.2 705.3 32.5 22.5 71.4 C

60 212.1 673.9 17.6 10.9 33.8 B

70 289.9 537.8 0.6 dominating 17.9 A


Intervention costs £334/year50 257.8 705.3 59.7 42.4 127.2 D

60 286.5 673.9 34.1 22.8 60.7 C

70 360.7 537.8 9.8 3.4 39.7 B



Results

96

then treatment from age 60 years was cost-effective(Figure 26).

AlfacalcidolThe meta-analysis of alfacalcidol RCTs suggested a marked decrease in vertebral fracture risk (RR = 0.46). The data were similar for hip fracture risk (RR = 0.25) and humeral fracture risk(RR = 0.19). However, the confidence estimateswere wide and the risk reduction was notsignificant for any of these fractures.

Cost-effectiveness improved with age but the cost-effective scenario was confined to women aged 70 years or more (Table 73a). When neutral effectson appendicular fractures were assumed, it was not cost-effective to treat at any age (Table 73b).

Incremental cost-effectivenessIncremental cost-effectiveness, using £30,000 cost per QALY gained, was undertaken for all cost-effective scenarios shown in Table 74 as grade B or better. Fluoride was not included

TABLE 72 Cost-effectiveness of using fluoride in women with established osteoporosis


Midpoint 90% CI

Results of meta-analysis50 155.4 700.2 dominated 20.8 dominated C

60 222.8 667.1 dominated 11.1 dominated C

70 486.7 523.3 dominated 4.2 dominated C

80 1012.0 357.3 dominated dominating dominated C

No effects on hip fracture50 138.3 704.0 30.8 26.9 36.6 C

60 180.8 672.1 17.4 14.7 21.5 A

70 324.0 532.9 13.5 11.2 16.9 A

80 665.4 369.7 6.8 4.1 11.8 A


100

80

60

40

20

00 10 20 30



25155

FIGURE 26 Distribution of cost-effectiveness of fluoride in women aged 60, 70 and 80 years with established osteoporosis (■, 60years; ●●, 70 years; ●, 80 years)


97

since it is not available in the UK and alfacalcidolwas also not considered since it is not licensed for use in osteoporosis in the UK. Treatments were ranked in ascending order of QALY gained. Marginal costs, QALYs and costs/QALY are shown in Table 75 for ages 50, 60, 70 and 80 years. At age 50 years, there were two cost-


TABLE 73 Cost-effectiveness of using alfacalcidol in women with established osteoporosis


Midpoint 90% CI

(a) Results of meta-analysis50 187.8 703.5 100.8 13.3 dominated C

60 229.0 671.7 41.6 4.7 dominated C



(b) No effects on appendicular fractures50 187.6 703.6 87.5 49.1 dominated D

60 229.9 671.9 39.0 25.8 dominated C

70 370.4 532.4 36.5 22.3 dominated C

80 713.0 369.0 30.1 4.6 dominated C


TABLE 74 Summary of grading of cost-effectiveness by agentand age

Agent Cost-effectiveness grade by age (years)

50 60 70 80

Raloxifenea D D B A

HRTb B B B B

Calcium C B B B

Calcium and vitamin D B A A A*

Calcitonin D D D C

Alendronate D C B A

Bisphosphonatec C B A A

Fluorided C A A A

Alfacalcidol C C B B

a Only when an effect on cardiovascular disease is assumed,plus no adverse effect on hip fractureb Adverse effects on breast cancer and beneficial effects oncardiovascular diseases are assumedc Price of £163/year; all osteoporotic patientsd Assumed not to affect appendicular fractures

TABLE 75 Marginal cost-effectiveness of interventionscompared with no treatment

Intervention Marginal cost Marginal(£000) QALY

Age 50 yearsHRT 48.4 1.65

Calcium and vitamin D 44.7 2.26

Age 60 yearsCalcium 45.4 2.13

HRT 43.6 3.53


Bisphosphonate 82.1 4.14


Raloxifene 140.1 3.01


HRT 28.4 5.63

Bisphosphonate 39.2 6.82

Alendronate 101.7 7.67


Raloxifene 108.6 7.13

Calcium and vitamin D –74.7 7.27

Bisphosphonate –55.1 8.66

HRT –12.4 9.69

Alendronate –12.2 9.73

Results

98

effective scenarios (calcium plus vitamin D,assuming effects on appendicular fractures and HRT in rank order). At age 60 years, the rank order of incremental cost-effectivenesswas calcium plus vitamin D, HRT, bisphosphonateand calcium. At age 70 years, the rank order was bisphosphonate, calcium plus vitamin D, HRT, alendronate, calcium and raloxifene. At 80 years, the rank order was bisphosphonate,alendronate, HRT, calcium plus vitamin D,raloxifene and calcium.

Sensitivity analysisAs discussed above, sensitivity analyses forvariations in effectiveness were included whenappropriate. Here, further analyses are confined to treatments that have been shown to be cost-effective, as shown in Table 74, unless otherwiseindicated. Important drivers of cost-effectivenessthat are evident from the preceding analysesinclude age and costs of the intervention.

AgeAge is clearly an important determinant of cost-effectiveness since the risk of fracturesincreases with age. It is clearly illustrated in the case of alendronate (see Table 69), for which the range of cost-effectiveness varied more than 60-fold between the ages of 50 and 80 years. It should be noted that improving cost-effectiveness with age is not invariant whenextraskeletal risks and benefits are included (e.g. raloxifene, see Table 63) or when neutraleffects on hip fracture are assumed (e.g. calcium, see Table 67a).

Costs of interventionAs expected, high costs of intervention areassociated with poorer cost-effectiveness since, in general, the variation in cost is greater than any proven variation in efficacy. This, however, is a generalisation: for example, the effects ofalendronate were better than those for fluoridedespite a sevenfold higher intervention cost (Table 76). This clearly rests on the assumptionsthat alendronate lowers the hip fracture rate(proven by RCT in established osteoporosis) and that fluoride does not (shown to increase hip fracture risk in epidemiological studies).

Confidence in mid-point estimatesThe sampling process using 1000 estimates of RRs permits an estimate of the 95% CI of allestimates for a given treatment at a given ageagainst a fixed estimate of no treatment. For all agents, the 95% CI lay within £1000/QALYgained at the thresholds of cost-effectiveness (data not shown).

Criteria for cost-effectivenessIn this study, the criteria for cost-effectiveness were based on cost-effectiveness being shown in 90% of cohorts. For example, grade A wasallocated to a treatment scenario in which, in 90% of the model runs, cost-effectiveness was less than £30,000/QALY gained (see Table 64).Decreasing the range to 80% of runs had a modest effect on the range of cost-effectiveness,particularly the lower estimate (Table 77). A greater effect was seen on the upper estimate, such that the grading of cost-effectiveness changedin a favourable manner. In the case of alendronateat the age of 70 years, grading would change

TABLE 76 Yearly costs and cost-effectiveness of different treatments

Agent Cost/year (£) Cost/QALY (£000)

Aged 70 years Aged 80 years

Calcitonin 2314 168 124

Alendronate 334 13.3 dominating

Raloxifenea 257 46.4 76.1

Bisphosphonate 163 5.7 dominating

Vitamin D derivativesa 157 36.5 30.1

HRTb 58 5.0 dominating

Calcium and vitamin D 55 dominating dominating

Fluoridea 48 13.5 6.8

Calcium 40 15.9 16.3

a Assumes for appendicular fractures that RR = 1b Assumes benefit in CHD and increased risk of breast cancer


99

from B (probably cost-effective) to A (always cost-effective). For calcium and vitamin D, a change of grade (from B to A) occurred at age 50 years.

Altering the threshold value for cost-effectivenessfrom £30,000 to £20,000 had a modest effect onthe grading of cost-effectiveness (Table 78). Adecrement in grading was observed for raloxifeneat ages 70 and 80 years, for HRT at age 50 years,for calcium and vitamin D at age 80 years, forcalcitonin at 80 years, for alendronate at 60 and 80 years, for bisphosphonate at 50 and 70 years, for fluoride at 50 and 60 years, and for alfacalcidol at 70 years.

DiscountingThe base case used a discount rate of 6% for costsand 1.5% for QALYs gained. The effect of discount-

ing the latter at 6% was significant. The effects of variations in discount rate for cost-effectivetreatments in women aged 60 years are shown in Table 79. The higher discount of benefits increasedthe cost-effectiveness ratio by approximately 25% at age 60 years. The higher discount value forQALYs increased cost-effectiveness by approxi-mately 63% at age 50 years and by 43% at age 70 years (data not shown). Thus, differences indiscount rates modify the conclusions concerningcost-effectiveness using the £30,000 threshold value.

CompliancePatients were said to be non-compliant if theyreceived 3 months of drug treatment and accrued no health benefit. With this definition,variations in compliance had a modest effect oncost-effectiveness, since costs as well as effectivenesschange in the same direction. The effect ofassuming 70% compliance was quantitatively muchless than for the variation in discount rate. Cost-effectiveness rose by less than 5% and did not alterthe overall conclusions on cost-effectiveness. ForHRT, cost-effectiveness rose by 1% or less assumingcompliance to be 70%, and by 5–10% assumingcompliance to be 30%. Further base-case examples are shown in Table 80.

Since the cost-effectiveness ratio increases withdecreasing compliance, an alternative way ofassessing the impact of non-compliance on cost-effective treatments is to determine at what level of compliance interventions no longer remain


TABLE 77 Range of cost-effectiveness estimates according topercentiles of the population used

Age (years) Cost-effectiveness estimate

90% CI 80% CI

Alendronatea

50 45.5 to 111.6 47.7 to 86.6

60 24.3 to 63.7 25.8 to 56.0

70 6.2 to 37.0 7.3 to 26.0

80 dominating to 21.3 dominating to 9.1

Bisphosphonatea

50 26.0 to 66.1 27.4 to 51.7

60 13.0 to 36.4 14.0 to 32.9

70 0.9 to 21.1 1.6 to 14.0


Calciuma

50 28.5 to 121.4 30.5 to 81.4

60 15.0 to 44.7 15.3 to 34.5

70 10.8 to 41.5 11.0 to 31.1

80 5.1 to 124.3 6.3 to 40.0

HRTa

50 30.9 to dominated 33.7 to dominated




Calcium + vitamin Da

50 14.2 to 33.6 14.9 to 28.2

60 6.0 to 16.1 6.5 to 13.6


80 dominating to dominating todominating dominating

a Base-case scenario

TABLE 78 Grading of cost-effectiveness using a threshold of£20,000 for cost/QALY (gradings using £30,000 threshold shownin parentheses if this differs from lower threshold)

Agent Cost-effectiveness grade by age (years)

50 60 70 80

Raloxifenea D D D (B) B (A)

HRTb C (B) B B B

Calciumc D (C) B B B

Calcium and vitamin D B A A A*

Calcitonin D D D D (C)

Alendronate D D (C) B B (A)

Bisphosphonate D (C) B B (A) A

Fluoridec D (C) B (A) A A

Alfacalcidol C C C (B) B

a Only when an effect on cardiovascular disease is assumedwith no adverse effect on hip fractureb When adverse effects on breast cancer and beneficialeffects on cardiovascular diseases are assumedc Assumed not to affect appendicular fractures

Results

100

cost-effective (i.e. exceed the £30,000/QALYthreshold). The compliance threshold is low, inthat a very significant majority of patients wouldneed to be non-compliant to make treatmentscenarios no longer cost-effective (Table 81). For example, if in women aged 80 years more than 5% were compliant, then treatment withalendronate would remain cost-effective within the limitations of the assumptions on compliance.

Offset timeIt was assumed in the base case that in mostinstances (but not calcitonin or calcium) theeffects of treatment wear off in a linear fashionover 5 years. Thus, 5 years of treatment incurssome benefit when treatment is stopped. Reducing the offset time to zero has a markedeffect on cost-effectiveness. In Table 82, threescenarios are shown for offset time followingbisphosphonate treatment in women.

It should be noted that these simulations wereundertaken on a single run of 8000 patients(rather than the smoothed approximation given by the Gaussian process), so that these centralpoint estimates differ somewhat from valuespreviously given and the errors (not given) are substantially greater.

Using a 10-year analytic time frame, cost/QALYgained increased at all ages when the offset timewas changed from 5 to 0 years.

In order to explore the effects of a longer offsettime, a 15-year analytic time frame was used. Asexpected, changing from a 5-year to a 10-yearoffset time improved cost-effectiveness. At age 50 years, the cost/QALY was above the £30,000threshold for cost-effectiveness but decreased from £49,000 to £25,100 with an offset time of 10 years.

Effect of changing T-scoreTreating women with a T-score lower than –2.5 SD had a very marked effect on cost-effectiveness (Table 83). For bisphosphonatetreatment, the intervention was not cost-effective atage 50 years. Increasing the stringency of the cut-off value for the T-score from –2.5 to –3.5 SD madetreatment cost-effective at age 50 years and costsavings were made at age 70 years or more.

Similarly, other treatments of borderline cost-effectiveness became cost-effective with increasingstringency of the T-score threshold. In women aged 70 years treated with raloxifene, the cost-effectiveness ratio fell from £26,500 at a T-score of –2.5 to £7538 at a T-score of –3.5 SD.

TABLE 79 Effect of variable discount rates for QALYs on cost-effectiveness at age 60 years

Agent Cost/QALY gained (£000) with QALY discounted at: Increment (%)

1.5% 6%

Alendronate 35.2 41.9 19

Bisphosphonatea 19.8 23.3 18

Bisphosphonateb 37.8 44.3 17

Bisphosphonatec 17.6 21.5 22

Bisphosphonated 34.1 41.7 22

Calcium 21.3 27.4 29

Calcium and vitamin D 9.2 11.7 27

HRTe 27.7 34.5 25

HRTf 12.7 16.6 31

HRTg 10.6 14.3 35

HRTh 12.4 16.6 34

a Aggregated effect, price of agent £163 p.a.b Aggregated effect, price of agent £334 p.a.c As a in patients with prior fracturesd As b in patients with prior fracturese Effects on vertebral fracturesf Effects on vertebral fracture and non-vertebral fractureg As e with additional effects on CHDh As f with additional effects on breast cancer and fixed effects on appendicular fractures


101

Costs of added years of lifeAs expected, the inclusion of future years ofmedical costs had effects on cost-effectivenessratios over the analytic time frame (10–15 years)but the effect at all ages was small and mostdeviations were the result of sampling errors (Table 84).

Duration of interventionCost-effectiveness improved as expected when the duration of treatment was increased from 5 to10 years (modelled in both cases over a 15-yearinterval). The effect can be judged indirectly bychanging the offset time from 5 to 10 years. Forbisphosphonate at age 50 years, the cost/QALYdecreased from £36,700 to £17,800 with theincreased offset time (see Table 84). At greater


TABLE 80 Cost-effectiveness (£000) of interventions accordingto age and compliance years

Age (years) Cost-effectiveness (£000)according to compliance

100 70 30

Alendronatea

50 61.4 62.8 68.6

60 35.3 36.1 39.7

70 13.3 13.7 15.8

80 –1.3b –0.9b 0.7

Bisphosphonatea

50 36.1 36.8 39.8

60 19.8 20.3 22.1

70 5.7 6.0 7.3

80 –6.4b –6.2b –5.3b

Calciuma

50 44.0 44.5 46.4

60 21.4 21.6 22.5

70 15.9 16.1 16.9

80 16.3 16.5 17.7

HRTa

50 61.9 62.7 65.9

60 27.8 28.1 29.5

70 24.1 24.4 25.9

80 19.9 20.2 21.7

Calcium and vitamin Da

50 19.8 20.1 21.2

60 9.2 9.4 10.1

70 0.5 0.6 1.1

80 –10.3 –10.2 –9.8

a Base-case scenariob Dominating

TABLE 81 Thresholds of compliance at which a cost-effectiveintervention exceeds a cost–utility of £30,000/QALY gained

Age (years) Cost/QALY gained Compliance(£000) threshold (%)

Alendronate70 13.3 5–10

80 –1.3 1–5

Bisphosphonatea

60 19.8 5–10

70 5.7 1–5

80 –6.4 1–5

Bisphosphonateb

60 17.6 5–10

70 0.6 1–5

80 –7.3 1–5

Calcium60 21.4 5–10

70 15.9 1–5

80 16.3 1–5

Calcium and vitamin Dc

50 19.8 5–10

60 9.2 1–5

70 0.5 < 1

80 –10.3 < 1

HRTd

60 27.8 20–30

70 24.1 10–20

80 19.9 5–10

HRTe

50 29.4 80–90

60 12.7 10–20

70 6.7 5–10

80 –1.9 1–5

HRTf

50 25.1 40–50

60 10.6 10–20

70 3.1 1–5

80 –3.0 1–5

HRTg

50 29.4 80–90

60 12.4 5–10

70 5.0 1–5

80 –1.3 < 1

a Aggregate of bisphosphonate data; costs £163 p.a.b Effectiveness in patients with prior fracturec Effectiveness on appendicular fractures (calcium with vitamin D)d Assumed effect on vertebral fracture alonee Effects on vertebral and non-vertebral fracturef As e but fixed effects on appendicular fractures and effectson CHDg As e but effects on CHD and breast cancer

Results

102

ages, the effect was smaller because of highermortality rates.

Mortality attributed to hip fractureIn the base case, it was assumed that 67% of alldeaths up to 90 days following hip fracture werecausally related to the hip fracture. If deaths be-yond 90 days were considered, the rate of causallyrelated deaths modelled fell to 42%. When alldeaths within 90 days were assumed to be causallyrelated (approximately 63% of deaths in the first

year), the cost-effectiveness ratio increased,although the effect was small and the overall con-clusions (i.e. grade of cost-effectiveness) did notmarkedly change. For example, treatment of apatient with bisphosphonate at age 60 years gave a cost-effectiveness ratio of £25,453. When alldeaths were assumed to be causally related, thecost-effectiveness improved to £22,392. The effectis likely to be less at younger ages since fewerwomen would sustain hip fractures and even fewer would be admitted to nursing homes.Conversely, the effect is more marked at olderages. At age 70 years, the cost-effectiveness ratiofell from £7710 to £5748, assuming the highermortality rate.

Effects on CHDBoth raloxifene and HRT may decrease the risk of CHD. These assumptions are based on epidemi-ological information in the case of HRT and onsurrogate markers in the case of raloxifene. The inclusion of these effects improves cost-effectiveness, particularly in the elderly, because of a marked increase in marginal QALY gained. In the case of both HRT and raloxifene, the cost-effectiveness improved in the elderly but remainedlittle changed at younger ages (Table 85).

Effect on breast cancerThe concern that HRT may increase the risk of breast cancer is largely derived from anepidemiological database. In this study, HRT was assumed variously to have no effect or to have an adverse effect on breast cancer. Theinclusion of an increase in risk of 35% over 5 years increased the marginal costs and decreased the QALYs gained. As for CHD, the effect was most marked in the elderly (see Table 85). Indeed, a 35% increase in breast cancer risk more or less negated a 34% decrease in CHD risk.

Prior fractureThe cohort modelled is a population of womenwith osteoporosis. Within this population, there is a mixed pattern of prior fragility fractures. The

TABLE 82 Effects of offset time on cost-effectiveness of abisphosphonate

Age (years) Offset Cost/QALYtime gained (£000)

(years)10-year 15-year

time frame time frame

50 0 71.05 36.1 49.0

10 25.1

60 0 22.95 19.8 15.3

10 16.7

70 0 12.15 5.7 3.2

10 1.7

80 0 –2.25 –6.4 –24.9

10 –19.2

TABLE 83 Effects of differences in BMD T-score on cost-effectiveness of bisphosphonate

Age Cost/QALY gained (£000)(years)

T-score = –2.5 SD T-score = –3.5 SD

50 36.1 13.9

60 19.8 8.2

70 5.7 –2.1

80 –6.4 –14.5

TABLE 84 The effect of adding cost of added years to cost-effectiveness of bisphosphonates

Age (years) 10-year analysis 15-year analysis

Base case + Cost of added years Base case + Cost of added years

50 36.1 39.1 49.0 48.5

60 19.8 22.5 15.3 16.5

70 5.7 9.4 3.2 6.9

80 –6.4 –8.7 –24.9 –7.2


103

distribution of fracture types is age dependent.Since different prior fractures have differentconsequences for further fracture, it is appro-priate to examine the effect of the type of fracture on treatment outcomes, viz.:

• prior forearm fracture• prior shoulder fracture• prior vertebral fracture• prior hip fracture.

The effect of bisphosphonate treatment in women at age 60 years with a BMD T-score of –2.5 SD in the presence of specific prior fracturesis shown in Table 86 compared with a base-casescenario (a population with a given distribution of prior fractures).

At this age, vertebral and forearm fractures are the most common, and the treatment of patientswith these fractures yields dividends in terms of

marginal QALYs greater than the base case. Thedividend is less than the base case for hip andshoulder fractures.

Criteria for vertebral fractureOur estimates suggested that vertebral fracturesthat come to clinical attention comprised approxi-mately 25% of all vertebral fractures. Clinicallycovert fractures diagnosed by vertebral morpho-metry are associated with significant morbidity and their exclusion will underestimate cost-effectiveness. In this sensitivity analysis, it wasassumed that the utility loss was half that of aclinically overt fracture, based on estimates ofhospital stay and changes in activities of dailyliving.343 If the utility loss is half that of a clinicallyovert fracture then, for every 100 fractures on X-ray, 25 will be clinically overt and the utility loss of 100 patients would be equivalent to 62.5 overtfractures, or 2.5 times greater than our estimates.

Increasing the apparent incidence of vertebralfracture by a factor of 2.5 improved cost-effectiveness, an effect more marked withincreasing age. At age 60 years, treatment withbisphosphonate decreased the cost-effectivenessratio from £25,453 to £21,548. At age 70 years, the ratio fell from £7710 to £775.

Clinical vignettes

From this review it can be seen that it is possible to deliver some treatments for established osteo-


TABLE 85 Effects on cost-effectiveness of changing assumptions relating to CHD and breast cancer risk

Age (years) RR reduction Marginal cost Marginal QALY Cost/QALY gained Grade(%) (£000) (£000)

Raloxifene and CHD50 0 140.2 0.94 148.5 D

80 0 108.4 1.43 76.1 D

50 20 140.2 0.95 148.1 D

80 20 108.6 7.13 15.2 A

HRT and CHD50 0 47.3 1.89 25.1 B

80 0 –33.8 6.98 –4.8 A*

50 34 47.3 1.89 25.1 B

80 34 –33.8 11.4 –3.0 A*

HRT, CHD (at +34%) and breast cancer50 0 48.4 1.65 29.4 B

80 0 –12.5 6.44 –1.9 A*

50 -35 48.4 1.65 29.4 C

80 -35 –12.4 9.69 –1.3 B

TABLE 86 Cost-effectiveness of bisphosphonate treatment inwomen aged 60 years according to site of prior fracture

Site of fracture Cost/QALY Marginalgained (£) QALY

Base case 19,842 4.14

Wrist 22,823 4.53

Proximal humerus 30,157 3.09

Spine 15,394 5.22

Hip 26,524 3.07

Results

104

porosis cost-effectively. The limitations of theanalyses for clinical care rest not only on the many assumptions to be made, but also on theanalytical approach. For example, as shown earlier,cost-effectiveness depends critically upon age, thetype of prior fracture and the T-score. Moreover,even though there was no RCT evidence for ahierarchy of treatment efficacy, differences in costand in apparent effectiveness, particularly on hipfracture risk, have marked implications for cost-effectiveness. Of particular therapeutic interest arethe effects of calcium, bisphosphonates and HRT,since these are widely used in the UK, and cost-effective scenarios are identified in this report.HRT is commonly targeted at the menopause,whereas calcium and vitamin D are mostcommonly used later in life in the elderly. The bisphosphonates are most widely used at intermediate ages, typically following a vertebral fracture.

Current practice guidelines adopted in the UKfollow a case-finding strategy in which potentialcandidates for treatment are identified by thepresence of strong risk factors for osteoporoticfracture. These risk factors include low body massindex, a family history of hip fracture and priorfragility fractures. These risk factors capture anelement of fracture risk over and above thatexplained by BMD. Treatment is indicated inindividuals subsequently found to have osteo-porosis. The present analysis provides a suitableframework for examining the cost-effectiveness of a major independent risk factor – namely, a priorfragility fracture. Common clinical situations aredescribed here briefly and questions raised relatingto the cost-effectiveness of given approaches againstthe background of current practice guidelines.

Forearm fractureForearm fractures commonly occur at the peri-menopause. Mrs X presented at age 50 years with aforearm fracture. She was at the peri-menopausebut had no menopausal symptoms. A subsequentBMD measurement identified her to be on thethreshold for osteoporosis. Can she be offered cost-effective treatment with HRT?

Inspection of Table 65 indicates that at age 50 years,treatment of patients with established osteoporosisis not cost-effective in the majority of women, usinga threshold value of £30,000/QALY gained. Themid-point estimate is £61,921 but falls within thethreshold to £29,399, assuming cardiovascular

benefits and breast cancer risks (see Table 66). Atthis age, the majority of the burden of osteoporosisis accounted for by fractures of the distal forearm.The mean cost-effectiveness ratio at this age forwomen with forearm fractures is £26,403, which isclose to the mid-point estimate of a population ofwomen at this age with established osteoporosisand a T-score of –2.5 SD. When the threshold forBMD is set at a T-score of –3.5 SD, treatmentbecomes very cost-effective, with a cost per QALYgained of £14,424. Thus there are cost-effectivescenarios to be found and, using more stringentcriteria than the WHO threshold for osteoporosis,treatment is very cost-effective.

Vertebral fractureThe mean age for vertebral fracture is in the mid-60s. Mrs Y sustained a vertebral fracture at age 60 years. A BMD examination showed her to have osteoporosis, with a T-score of –3.0 SD. Canshe be given a bisphosphonate (etidronate oralendronate) cost-effectively?

The cost-effectiveness ratio with bisphosphonate at age 60 years at the threshold of osteoporosis (T-score = –2.5) is £36,000 but decreases to £6100 with a T-score of –3.5 (see Table 83). In the presence of a prior vertebral fracture and a T-score of –3.0 SD, the intermediate ratio is£17,584. The value lies within the threshold forcost-effectiveness. Note that the combinedbisphosphonate effect is modelled on the price of etidronate (£163 annually). When the highertreatment cost of alendronate is used, the cost-effectiveness ratio increases to £31,178.

Hip fractureThe average age for hip fracture in NorthernEurope is 80 years. Mrs Z lived alone and sustaineda hip fracture at age 85 years. Could she be treatedcost-effectively with calcium and vitamin D withoutrecourse to a BMD measurement?

Here it is assumed that calcium plus vitamin Ddecreases the risk of appendicular fractures to anextent described for women in sheltered accommo-dation. In this scenario, treatment of all womenwith established osteoporosis is cost saving at age80 years (see Table 67). The average T-score ofwomen aged 85 years is –2.5 SD so that it will, onaverage, also be cost-effective to intervene, evenwithout the assessment of BMD. Note that at thisage, bisphosphonate and alendronate are evenmore cost-effective.


105

Acost-effectiveness analysis was undertaken of interventions in established osteoporosis.

The approach was to review systematically theevidence for efficacy from RCTs, costs and health-state utility values. A model was constructed thatwas populated with hazard functions drawn when-ever possible from the UK. A novel feature of themodel is that ranges of cost-effectiveness could bedetermined which take account of the uncertain-ties surrounding the effectiveness of intervention.

The principal findings show that there are effectivetreatments for established osteoporosis in womenand that some of these treatments can be givencost-effectively. However, not all treatments wereshown to be effective and, even when efficacy was demonstrated, they were not invariably cost-effective. No single agent was cost-effective over the entire age range relevant for post-menopausalosteoporosis, with exception of HRT and thepossible exception of calcium (with vitamin D), as shown in Table 74.

The incremental cost-effectiveness suggests thatcalcium is the agent of choice at ages 50–70 yearsand bisphosphonate at age 80 years. There are,however, difficulties in assigning a rank order ofpreference. First, the meta-analysis of effectivenesscould not distinguish significant differences ineffectiveness between agents, although the mid-point estimates vary. Second, untested assumptionswere made for efficacy. For example, calcium andvitamin D were assumed to decrease appendicularfractures in the general population (as shown byan RCT for women in sheltered accommodation).Moreover, the efficacy has not been tested inwomen aged 50 years. The effectiveness ofbisphosphonate is an aggregate effect but theintervention cost was modelled for etidronate.These various considerations suggest that it would be unwise at present to recommend ahierarchy of preferred treatments.

The observation that most interventions are notcost-effective at all ages indicates that conclusionsare very sensitive to the assumptions used topopulate the model and to the modelling tech-nique. Many of these assumptions have beenidentified; only those of particular importance to the conclusions or recommendations are

reviewed below. However, the vast majority of theassumptions used are conservative. This, in turn,should modulate the interpretation of healtheconomics analyses, in the sense that scenarios that demonstrate cost-effectiveness are likely to be robust but scenarios without, or borderline,cost-effectiveness may well be cost-effective but are surrounded by uncertainty. Moreover, lack of cost-effectiveness is not the sole arbiter ofclinical utility. Assumptions giving rise touncertainties of major importance include:

(a) treatment effects(b) health states in established osteoporosis(c) the hazard function in established

osteoporosis(d) the constraints of the model.

Treatment effects

Hip fracture accounts for the greatest morbidityand costs of osteoporosis, and drives health eco-nomics considerations. In the absence of anyassumed effect on hip fracture, cost-effectivescenarios using base-case assumptions from RCTsare confined to the very elderly (e.g. HRT, calcium plus vitamin D). There is a paucity ofavailable information from RCTs on hip fractureoutcomes in established osteoporosis. Indeed, at the cut-off date for this analysis, only foralendronate had a significant effect on hip fracture rates been demonstrated by RCT. Therewas information from RCTs on hip fracture withthe use of vitamin D with or without calcium and hip protectors, but these studies did notspecifically examine women with osteoporosis. For this reason, such studies were excluded fromthe meta-analysis of efficacy. Moreover, there were epidemiological data suggesting that anumber of other interventions decrease the risk of hip fracture. These included HRT, calcitonin,anabolic steroids, thiazide diuretics and etidronate.Because these findings were unsupported by RCTs, they were not considered in the meta-analysis but were included in sensitivity analyses of cost-effectiveness when beneficial effects onother fracture outcomes had been proven. In the summary of cost-effectiveness, some of these assumptions are embraced.


Chapter 7

Discussion and conclusions


106

The logic of this approach is that osteoporosis is asystemic disease affecting all regions of the skeleton.It may, therefore, be counter-intuitive to assume that,when agents acting systematically have demonstratedefficacy on vertebral fracture, there is no effect onhip fracture risk, particularly when this is supportedby observational studies. On the other hand, pre-cedents discussed in chapter 2 indicate that cautionis required, in that RCTs suggest that raloxifene andfluoride have little or no effect on hip fracture risk,despite RCT evidence for significant effects onvertebral fracture risk. In such cases, a neutral effect (RR = 1.0) has been conservatively assumed in the sensitivity analyses.

As mentioned, the absence of RCT data on hipfracture outcome should not necessarily be takento mean that agents do not favourably affect hipfracture risk. Indeed, the balance of probabilitiessuggests otherwise. There is also a lack of con-vincing RCT data about the effects of HRT orraloxifene on cardiovascular outcomes. Theseabsences limit interpretation of the data. When hipfracture effects have been assumed from epidemi-ological data, it is important to acknowledge thatthe quantum of effect is uncertain, due to intrinsicpopulation biases that are unquantifiable. A goodexample is HRT, for which large effects on hipfracture risk are shown in many observationalstudies but the only RCT (albeit in patients withoutosteoporosis) showed no significant effect. Forthese reasons, the hip fracture effect used in thesensitivity analysis was modest. These consider-ations suggest that many of the conclusionsrelating to efficacy will be conservative.

A further problem relating to the treatmentsidentified here is that the responses to interventionin terms of vertebral fracture outcome may be non-linear, as the greatest risk reduction is seen in theearly years of intervention. If true, this suggeststhat for some treatments, the longer the durationof treatment, the lower the RR reduction. Untilthese uncertainties are resolved, analyses of treat-ments for more than 5 years become progressivelymore speculative; this is the principal reason whythe treatment time frame was restricted to 5 years.Conversely, however, shorter intervention timesmight usefully be modelled when more infor-mation becomes available on offset times.

There is good evidence that the offset of effect of intervention is not instantaneous. Offset timeshave not been systematically studied, although they form a recommendation by WHO in drugdevelopment,50 a view that is endorsed. A review of the available evidence suggests that the chosen

3–5 year offset time might be conservative for some agents. The offset time has been shown to be a critical component of apparent cost-effectiveness.

Also, the vast majority of studies compared the test agent plus calcium and/or vitamin D with aplacebo with calcium and/or vitamin D. In effect,these are trials of superiority. On the assumptionthat calcium with or without vitamin D has intrinsictherapeutic effects, then the efficacy of the testagent may be underestimated. Such an assumptionwould only be valid if it could be shown that theeffect observed with combination treatment wasgreater that the effect of the test agent alone. Such data are not available. Moreover, the effects of calcium and/or vitamin D appear to be greater in individuals with a poorer nutritionalstatus, so that in RCTs of new agents, any effect of calcium and vitamin D alone is uncertain. Forthese reasons, the observed effects of test agentswere not adjusted to take account of potentialeffects of calcium and vitamin D, although it is acknowledged that this may be a con-servative position.

There are additional considerations for several ofthe therapeutic modalities discussed and for whichcost-effectiveness has not been demonstrated.

1. Raloxifene has not been shown to be cost-effective below the age of 60 years (even whenbeneficial effects on CHD are assumed).The major reason for this conclusion is theabsence of effect on the risk of appendicularfractures. There were no RCTs that examinednon-vertebral fractures as a primary endpoint.Direct information on cardiovascular outcomesis likely to be available shortly, which mightsignificantly temper our conclusions. Sinceraloxifene is generally not cost-effective inestablished osteoporosis, it will not be cost-effective in the prevention of osteoporosis. It is relevant that the time frame modelled was 10 years and, if longer-term effects on breast cancer mortality were assumed, it is likely that cost-effective scenarios at youngerages would be found.

2. Calcitonin has not been shown to be cost-effective at any age.The cost of calcitonin rather than its efficacy isthe principal determinant of this outcome. Itmay be relevant that intranasal calcitonin hasjust been approved in the UK and, from pre-cedents elsewhere in Europe, may have a morefavourable pricing structure. If so, then furtheranalyses would be of considerable interest.


107

Health state utility values inestablished osteoporosisThe systematic review has highlighted the paucityof data available on health state utility values inestablished osteoporosis. The health state valuesfor each fracture are based on a critical review ofthe literature but, clearly, further work needs to bedone in this area, using standardised methodologyover the different health states. Since the comple-tion of this study, further empirical data have be-come available on health state values in establishedosteoporosis in patients referred to a hospitalaccident department in Malmö, Sweden.344 If suchdata are applicable to all patients with symptomaticvertebral fractures, they suggest that the disutilityused here might be doubled and substantially alter the conclusions on cost-effectiveness.

There is a particular problem in patients withestablished osteoporosis. Such individuals havealready sustained a fragility fracture. Thus, the rele-vant question is the impact of a second fracture onexisting health states – and there is absolutely noliterature available on this. This has posed problemsin modelling. It has been assumed that a secondfracture at a different site will incur disutility attri-butable to both fractures using a multiplier. Forexample, an individual with a Colles’ fracture whosustained a hip fracture would be assigned a utilityvalue equivalent to that of a hip fracture multipliedby the remaining disutility of the Colles’ fracture. It has also been assumed that an individual with aprior hip fracture would incur no further morbidityif a second fracture was sustained, other than in thefirst year of the second fracture. Similarly, a womanwith a vertebral fracture would sustain no furtherdisutility from a second vertebral fracture other than in the first year. This may well be a con-servative scenario and, again, underestimate the utility losses in osteoporosis.

It has not been possible to quantify the impact ofside-effects on cost–utility apart from in CHD andbreast cancer. Many of the agents evaluated have theadditional unwanted effects identified in this review.The health state consequences are unknown and arenot considered here. This is an important omissionsince risks that have effects on the treated populationas a whole are likely to affect cost–utility markedly.

Hazard functions in establishedosteoporosisThe principal modelling exercise was under-taken in cohorts of women at the boundary

of osteoporosis (i.e. with a T-score of –2.5 SD). In practice, few individuals diagnosed as havingestablished osteoporosis would have a T-score of exactly this value. Indeed, distribution theoryindicates that the RR for patients with osteo-porosis is approximately twice that of individuals at the threshold for osteoporosis (see Table 40).The excess risk is approximately equivalent to a T-score of –3.5 SD. As shown here, cost-effectiveness is critically dependent on the T-score.

With respect to outcomes, the major uncertainty in the analysis related to vertebral fracture risk,with its problematic epidemiology. Uncertaintiesarose because of multiple methods of diagnosisand their uncertain relevance to patients. In thisstudy, the incidence of vertebral fractures thatcome to clinical attention has been computedusing data derived from Sweden rather than the UK. Although the estimates are likely to bereasonably accurate, there is increasing evidencethat vertebral deformities that do not present to clinical attention occasion a substantialmorbidity. They also presage future fractures. As shown by the sensitivity analysis, this exclu-sion has a substantial impact on interventionthresholds based on cost-effectiveness. Against this background, the approach used is conservative.

Constraints of the model

Multiple outcomes are a major problem in theevaluation of osteoporosis. These include not only the many different fractures, each with differ-ent morbid consequences, but also non-skeletaloutcomes. There is, for example, evidence that the risk of breast cancer is lower in women withosteoporosis than in those without. Moreover,several interventions may affect non-skeletal out-comes. Classic examples are HRT and raloxifene,both of which may affect the risk of breast cancerand CHD. One of the strengths of the modellingapproach is that it can accommodate, on a patient-by-patient basis, the multiple transitions that thedisorder demands. It also indicates the need to use a cost–utility approach.

Although the model is very flexible and can handle many transition probabilities, it was im-peded by a lack of information on many of thetransitions that arise in such a complex disorderwith multiple outcomes. Whenever possible, con-servative assumptions were used. An exception was the relationship between BMD and fracture



108

risk. It was assumed that the average risk offracture (i.e. in the general population) occurswith an average BMD. This is a simplificationbecause BMD is normally distributed, whereas the risk of many fractures, hip fracture in partic-ular, increases exponentially with decreasing BMD. Thus, a given BMD will overestimate the risk of fracture when the fracture hazard is derivedfrom the average fracture rate. However, not allosteoporotic fractures (although a substantialmajority) were accommodated, largely because of the lack of relevant information to populatethese transition states. This is clearly an area for further research.

Although the first phase of modelling was based on 8000 patients for each set of parameters, thismay not eliminate noise for infrequent events. For example, the probability of some events, such as hip fracture and entry to nursing homefollowing a hip fracture is low, being less than 0.6% and 0.04%, respectively, at age 60 years. It is likely that more than 8000 patients would be required to significantly reduce noise. Withsuch small values, the reduction in events becauseof treatment, assuming RR = 0.5 for hip fracture, is small compared with the probability of deathdue to natural causes (approximately 0.9% at age 60 years).

Implications for practice

The results presented in chapter 6 and the summaryof findings presented in Table 74 provide importantinformation for policy makers on the cost-effectivetreatment of established osteoporosis. The principalfindings are that there are effective treatments forestablished osteoporosis in women and that some of these can be cost-effective. However, not all treat-ments were shown to be effective and, even whenefficacy was shown, they were not invariably cost-effective. Few agents are cost-effective over theentire age range relevant for post-menopausalosteoporosis, as shown in Table 74.

However, it was not possible to identify withconfidence a hierarchy of interventions based oneffectiveness. An incremental cost-effectivenessanalysis by age group was undertaken but greatcaution is required in suggesting the order inwhich treatments should be considered.

Important areas of further research (outlinedbelow) have been identified that will, when theresults become available, inform policy moreaccurately in the future.

Recommendations for further researchThere are several acknowledged deficiencies in this study that form the basis of theserecommendations.

1. Intervention thresholds. The acceptedoperational definition of osteoporosis rests upon the measurement of BMD at the hip.Osteoporosis is defined as a T-score of –2.5 SD or less. The widespread acceptance of thiscriterion has meant that T-score thresholds have been used for drug development, as well as for practice guidelines.9,141,345 It is evident,however, that the same T-score has differentsignificance at different ages and in differentclinical contexts. As shown here, the presence of a prior fracture increases fracture risk overand above that accounted for by BMD, and the risk of fracture in the absence of a priorfracture depends critically upon age. Forexample, the incidence of hip fracture inwomen from the UK with a T-score of –2.5 SDvaries from 0.25% at ages 50–54 years to 3.2% at age 90 years or more (see Table 38). It isevident, therefore, that diagnostic thresholdsshould differ from intervention thresholds, even without health economics considerations.Indeed, it is the view of both WHO and theInternational Osteoporosis Foundation that, in the future, intervention thresholds should be based upon absolute fracture probabilities,such as 10-year risk.346 As assessment guidelinesdevelop, it will be important to change theanalytical framework of health economicsassessments to accommodate these concepts.

2. Adverse effects. Because of lack of resources, the impact of adverse effects other than breastcancer or CHD has not been considered. How-ever, common adverse effects in the treatedpopulation would have a marked negative effect on cost-effectiveness. A systematic reviewof side-effects and their associated utility stateswould determine whether these might beincluded in further health economics analysis.

3. Prior fractures. The cohorts of women modelledhad a range of osteoporotic fractures. However,different prior fractures have different futureconsequences, as shown in the sensitivity ana-lysis. For example, treatment of patients with a prior vertebral fracture is more worthwhilethan treatment of women with a prior forearmfracture. The analysis of this is not exhaustiveand is amenable to further research.

4. Health state utility values. There is a dearth of empirical data on these values in established


109

osteoporosis using standardised methodology. Ofparticular concern is that the health utility statesin second and subsequent years may have beenseriously mis-estimated. Again, this is bounded by the lack of empirical data and hence is recom-mended as an area for further research. This will require the administration of standardised,preference-based, generic measurements ofhealth status to a large prospective populationcohort with long-term follow-up.

5. Symptomatic vertebral fractures. There is increas-ing evidence that these are associated with signifi-cant mortality. In addition, it is likely that healthutility states are underestimated. These deficien-cies will affect conclusions concerning cost-effectiveness. However, the consideration of mor-phometric deformities and attendant morbidity is likely to have a much greater impact. Moreprecise information on the incidence of vertebraldeformities and their associated impact on

quality of life is likely to become available within the next year, and its inclusion in health eco-nomics models will be an important area forfurther research.

6. Costs and effectiveness of interventions. Theevidence base for these is changing rapidly. New agents are undergoing clinical develop-ment, for example, parathyroid hormone,risedronate and other SERMs. In addition, new formulations are now available in the UKwith lower intervention costs than those usedhere, such as the oestrogens, alendronate, andcalcium with vitamin D. Since the completion of this data analysis, intranasal calcitonin hasbecome available in the UK and is likely to bepriced more competitively than other forms.Outcomes on hip fracture are expected fromintervention with vitamin D, calcium and HRT.As further information become available, it willbe important to update the present analysis.



111

This report was commissioned by the NHS R&D HTA Programme (project number

95/11/04). The views expressed in this report arethose of the authors and not necessarily those ofthe NHS R&D HTA Programme. Any errors arethe responsibility of the authors.

The authors are grateful to Professor David Barlow(John Radcliffe Hospital, Oxford), Professor Cyrus

Cooper (Southampton General Hospital,Southampton) and Dr Eugene McCloskey (WHOCollaborating Centre for Metabolic Bone Disease,Sheffield) for acting as consultants during this work.

The authors are also indebted to the referees fortheir perseverance in reading the report and thequality of their comments.


Acknowledgements


113

1. Office of Technology Assessment. Effectiveness and costs of osteoporosis screening and hormonereplacement therapy. Pittsburgh: Office ofTechnology Assessment; 1995.

2. Daly E, Roche M, Barlow D, Gray A, McPherson K,Vessey M. HRT: an analysis of benefits, risks andcosts. Br Med Bull 1992;48:368–400.

3. Pitt FA, Lloyd Jones M, Brazier JE, McGrother CW,Kanis JA, Wallace WA, et al. The costs and benefitsof screening and preventing post-menopausalosteoporosis in Trent RHA. Sheffield: TrentOsteoporosis Working Group; 1990.

4. Pitt FA, Brazier JE, Kanis JA, Wallace WA,McGrother C, Radley H, et al. Guidance for theprevention of osteoporosis in general practice.Sheffield: Trent Osteoporosis Working Group; 1991.

5. Whittington R, Faulds D. Hormone replacementtherapy: 1. Pharmacoeconomic appraisal of itstherapeutic use in menopausal symptoms andurogenital oestrogen deficiency. Pharmacoeconomics1994;5:419–45.

6. Whittington R, Faulds D. Hormone replacementtherapy: 2. Pharmacoeconomic appraisal of its rolein the prevention of postmenopausal osteoporosisand ischaemic heart disease. Pharmacoeconomics1994;5:513–54.

7. Weinstein MC. Estrogen use in postmenopausalwomen – costs, risks and benefits. N Eng J Med1980;6:308–16.

8. Weinstein MC, Tosteson ANA. Cost-effectiveness of hormone replacement. Ann NY Acad Sci1990;592:162–72,185–9.

9. National Osteoporosis Foundation. Osteoporosis:review of the evidence for prevention, diagnosisand treatment and cost-effectiveness analysis. Statusreport. Osteoporos Int 1998;8 Suppl 4:1–88.

10. Jonsson B, Christiansen C, Johnell O, Hedbrandt J.Cost effectiveness of fracture prevention in estab-lished osteoporosis. Osteoporos Int 1995;5:136–42.

11. Torgerson DJ, Kanis JA. Cost-effectiveness ofpreventing hip fractures in the elderly populationusing vitamin D and calcium. QJM 1995;88:135–8.

12. Consensus Development Conference. Diagnosis,prophylaxis and treatment of osteoporosis. Am JMed 1991;90:107–10.

13. World Health Organization. Assessment of fracture risk and its application to screening forpostmenopausal osteoporosis. Technical ReportSeries no 843. Geneva: WHO; 1994.

14. Kanis JA, Melton LJ, Christiansen C, Johnston C,Khaltaev A. The diagnosis of osteoporosis. J BoneMiner Res 1994;9:1137–41.

15. Kanis JA, Gluer CC. An update on the diagnosisand assessment of osteoporosis with densitometry.Osteoporos Int 2000;11:192–202.

16. Marshall D, Johnell O, Wedel H. Meta-analysis ofhow well measures of bone mineral density predictoccurrence of osteoporotic fractures. BMJ1996;312:1254–9.

17. Kanis JA, Pitt F. Epidemiology of osteoporosis. Bone 1992;1:S7–15.

18. Report on osteoporosis in the EuropeanCommunity. Action for prevention. Luxembourg.Office for Official Publications of the EuropeanCommunities; 1998.

19. Kanis JA. Textbook of osteoporosis. Oxford:Blackwell Science; 1996.

20. Dolan P, Torgeson DJ. The cost of treatingosteoporotic fractures in the United Kingdomfemale population. Osteoporos Int 1998;8:611–17.

21. Melton LJ, Kan SH, Wahner WA, Riggs BL. Lifetimefracture risk: an approach to hip fracture riskassessment based on bone mineral density and age. J Clin Epidemiol 1988;41:989–94.

22. Oden A, Dawson A, Dere W, Johnell O, Jonsson B, Kanis JA. Lifetime risk of hip fracture is underestimated. Osteoporos Int 1998;8:599–603.

23. Lauritzen JB, Shorts P, McNair P, Lurid B, Transbol I. Radial and femoral fractures aspredictors of subsequent hip, radial or humeralfractures, and their seasonal variations. OsteoporosInt 1993;3:133–7.

24. Kanis JA, Johnell O, Oden A, Sernbo I, Redlund-Johnell I, Dawson A, et al. Long term risk ofosteoporotic fracture in Malmo. Osteoporos Int2000;11:669–74.

25. Singer BR, McLauchlan GJ, Robinson CM, Christie J. Epidemiology of fractures in 15,000adults. J Bone Joint Surg Br 1998;80:243–8.

26. Ellfors L, Allander E, Kanis JA, Gullberg B, Johnell O, Dequeker A, et al. The variable incidenceof hip fracture in southern Europe: the MEDOSstudy. Osteoporos Int 1994;4:253–63.


References

References

114

27. Melton LJ, Chrischilles EA, Cooper C, Lane AW,Riggs BL. How many women have osteoporosis? J Bone Miner Res 1992;7:1005–10.

28. Melton LJ, O’Fallon WIM, Riggs BL. Secular trendsin the incidence of hip fractures. Calcif Tissue Int1987;41:57–64.

29. Spector TD, Cooper C, Lewis AF. Trends inadmission for hip fracture in England and Wales1968–1985. BMJ 1990;3:1173–4.

30. Lancaster HO. Expectations of life. Berlin:Springer-Verlag; 1990. p.33.

31. Hoffenberg R, Jarnes OFW, Brocklehurst JC, Green ID, Horrocks P, Kanis JA, et al. Fracturedneck of femur. Prevention and management. J R Coll Physicians Lond 1989;23:8–12.

32. Cooper C, Campion C, Melton LJ. Hip fractures inthe elderly: world-wide projections. Osteoporos Int1992;2:285–9.

33. Gullberg B, Johnell O, Kanis JA. Worldwideprojection for hip fracture. Osteoporos Int1997;7:407–13.

34. Riggs BL, Melton LJ. The prevention and treatmentof osteoporosis. N Eng J Med 1992;327:620–7.

35. Chapuy MC, Arlot ME, Delmas PD, Meunier PJ.Effect of cholecalciferol treatment for three yearson hip fracture in elderly women. BMJ1994;308:1081–2.

36. Kannus P, Parkkari J, Pasanen SNM, Palvanen M,Jarvinen M, Vuori L. Prevention of hip fracture inelderly people with use of a hip protector. N Eng JMed 2000;343:1506–13.

37. Kanis JA. The use of calcium in the management of osteoporosis. Bone 1999;24:279–90.

38. Johnell O, Gullberg B, Kanis JA, Allander E, Elffors L, Dequeker J, et al. Risk factors for hipfracture in European women: the MEDOS study. J Bone Miner Res 1995;10:1802–15.

39. Kanis, JA. Screening of post-menopausalosteoporosis. A review. London: Department ofHealth; 1992.

40. Osteoporosis after fractures. Clinical guidelines for management amongst older people. London:Royal College of Physicians of London; 1996.

41. Kanis JA, Johnell O, Oden A, de Laet C, Jonsson B,Dawson A. Ten year risk of osteoporosis fractureand the effect of risk factors on screening strategies.Bone 2001;30:251–8.

42. Kanis JA, Johnell O, Oden A, Jonsson B, de Laet C,Dawson A. Prediction of fracture from low bonemineral density overestimates risk. Bone2000;26:387–91.

43. Kanis JA. Treatment of osteoporosis in elderlywomen. Am J Med 1995;98 Suppl 2A:S60–6.

44. Jonsson B, Kanis JA, Dawson A, Oden A, Johnell O.Effect and offset of effect of treatments for hipfracture on health outcomes. Osteoporos Int1999;10:193–9.

45. Barlow D. Report of Department of Health AdvisoryGroup on Osteoporosis. London: Department ofHealth; 1994.

46. Compston JE, Cooper C, Kanis JA. Bone densito-metry in clinical practice. BMJ 1995;310:1507–10.

47. Kanis JA, Geusens P, Christiansen C. Guidelines forclinical trials in osteoporosis. A position paper forthe European Foundation for Osteoporosis andBone Disease. Osteoporos Int 1991;1:182–8.

48. Note for guidance on involutional osteoporosis inwomen. CMPM/EWP/552/95. London: Committeefor Proprietary Medicinal Products; 1997.

49. Guidelines for preclinical and clinical evaluation of agents used in the prevention or treatment ofpostmenopausal osteoporosis. Rockville, MD: Foodand Drug Administration, Division of Metabolismand Endocrine Drug Products; 1994.

50. World Health Organization. Guidelines forpreclinical evaluation and clinical trials inosteoporosis. Geneva: WHO; 1998.

51. Tosteson AN, Rosenthal DI, Melton LJ 3rd,Weinstein MC. Cost effectiveness of screeningperimenopausal white women for osteoporosis:bone densitometry and hormone replacementtherapy. Ann Intern Med 1990;113:594–603.

52. Kanis JA. What constitutes evidence for drugefficacy in osteoporosis? Drugs Aging 1993;3:391–9.

53. Cummings SR, Black DM, Vogt TM. Changes inBMD substantially underestimate the anti-fractureeffects of alendronate and other antiresorptiveagents. J Bone Miner Res 1996;11:S102.

54. van Staa TP, Abenhaim L, Cooper C. Use of cyclical etidronate and prevention of non-vertebralfractures. Br J Rheumatol 1998;37:87–94.

55. Kanis JA, Johnell O, Gullberg B, Allander E, Dilsen G, Gennari C, et al. Evidence for efficacy ofdrugs affecting bone metabolism in preventing hipfracture. BMJ 1992;305:1124–8.

56. Shiraki M. [Vitamin K2. In Japanese] Nippon Rinsho1998;56:1525–30.

57. Nakatsuka K, Inaba M, Aratani H, Iba K, Sato T,Koike T, et al. [Effects of long-term administrationof alfacalcidol on bone mass and bone metabolismin patients with primary osteoporosis – comparisonwith calcium preparations. In Japanese.] NipponRonen Igakkai Zasshi 1997;34:569–76.

58. Fujita T, Orimo H, Inoue T, Kaneda K, Sakurai M,Morita R, et al. [Double-blind multicentre com-parative study with alfacalcidol of etidronatedisodium (EHDP) in involutional osteoporosis. In Japanese] Rinsho Hyoka 1993;21:261–302.


115

59. Overgaard K, Ravn P, Hansen MA, Christiansen C.[Salmon calcitonin in osteoporosis. The effect ofintranasal application on bone mineral content andfracture frequency in postmenopausal women withmanifest osteoporotic changes. In Dutch.] UgeskrLaeger 1993;155:2387–91.

60. Overgaard K, Hansen MA, Jensen SB, Christiansen C. Effect of salcatonin givenintranasally on bone mass and fracture rates inestablished osteoporosis: a dose–response study.BMJ 1992;305:556–61.

61. Agrawal R, Wallach S, Cohn S, Tessier M, Verch R,Hussain M, et al. Calcitonin treatment of osteo-porosis. In: Calcitonin 1980: proceedings of aninternational symposium; Milan, October 1980.Amsterdam: Excerpta Medica; 1981 p.237–46.

62. Buckle RM. Three year study of sodium fluoridetreatment on vertebral fracture incidence inosteoporosis. J Bone Miner Res 1989;4:S186.

63. Dykman TR, Haralson KM, Gluck OS, Murphy WA,Teitelbaum SL, Hahn TJ, et al. Effect of oral 1,25-dihydroxyvitamin D and calcium on glucocorticoid-induced osteopenia in patients with rheumaticdiseases. Arthritis Rheum 1984;27:1336–43.

64. Fujita T, Fukase M, Shimada T, Yamamoto H.Treatment of established osteoporosis with 1 alpha (OH) vitamin D3 and low dose intermittentelcatonin (eel calcitonin derivative). J Bone MinerMetab 1992;10:37–40.

65. Gennari C, Chierichetti SM, Bigazzi S, Fusi L,Gonnelli S, Ferrara R, et al. Comparative effects onbone mineral content of calcium and calcium plussalmon calcitonin given in two different regimes inpostmenopausal osteoporosis. Curr Therapeut Res1985;38:455–64.

66. Hayashi Y. Investigation of fracture frequency ofJapanese osteoporosis patients: effects of drugtherapy. J Bone Miner Metab 1988;6:120.

67. Hayashi Y, Fujita T, Inoue T. Decrease of vertebralfracture in osteoporotics by administration of 1 alpha-hydroxy-vitamin D3. J Bone Miner Metab1992;10:184–8.

68. Shiraki M, Shiraki Y, Aoki C, Miura M. Vitamin K2 (menatetrenone) effectively prevents fracturesand sustains lumbar bone mineral density inosteoporosis. J Bone Miner Res 2000;15:515–21.

69. Dilsen G, Gulbaba G, Sindel D. Calcitriol in thetreatment of osteoporosis. Osteoporos Int 1998;8:101.

70. Ebeling PR, Wark JD, Yeung S, Poon C, Salehi N,Nicholson GC, et al. Effects of calcitriol or calciumon bone mineral density and fractures in men with idiopathic osteoporosis. Osteoporos Int 1998;8 Suppl 3:114.

71. Gallagher JC. Treatment of postmenopausalosteoporosis with fluoride plus either calcium or calcitriol. J Bone Miner Res 1992;7:S318.

72. Rozhinskaya L, Marova E, Sazonova N. Effectiveness of monofluorophosphate inestablished steroid osteoporosis. Osteoporos Int1999;9(4 Suppl 1):S11–12.

73. Brockstedt H, Melsen F, Mosekilde L. Double-blind,randomized study comparing the therapeuticeffects of continuous and pulse dosing sodiummonofluorophosphate (NaMFPh) in spinal crushfracture osteoporosis [abstract]. Osteoporos Int1996;6 Suppl 1:254.

74. Gallagher JC. Effect of fluoride in osteoporosis[abstract]. J Bone Miner Res 1990;5 Suppl 1:S230.

75. Guañabens N, Farrerons J, Perez-Edo L, Monegal A, Renau A, Roca M, et al. Comparison of the efficiency and safety of cyclic etidronateversus sodium fluoride in postmenopausalosteoporosis [abstract]. Osteoporos Int 1996;6:255.

76. Leidig-Bruckner G, Eberwein S, Bruckner T, Wolf S, Ziegler R. Long term follow-up of lumbarand femoral bone mineral density and vertebralfractures in patients with manifest postmenopausalosteoporosis after fluoride therapy alone orcombined with estrogens or anabolica [abstract].Exp Clin Endocrinol Diabetes 1997;105:A58.

77. McClung M, Bensen W, Bolognese M, Bonnick S,Ettinger M, Harris S, et al. Risedronate increasesbone mineral density at the hip, spine and radius in postmenopausal women with low bone mass.Osteoporos Int 1998;8:111.

78. Chesnut CH 3rd, Baylink DJ, Doyle D, Genant H,Harris S, Kiel DP, et al. Salmon calcitonin nasalspray prevents vertebral fractures in establishedosteoporosis: further interim results of the“PROOF” study. Osteoporos Int 1998;8 Suppl 3:13.

79. Gutteridge DH, Drury PH, Stewart GO, Prince RL,Price RI, Retallack RW, et al. Cyclic Na fluoride +HRT in postmenopausal osteoporosis – morevertebral fractures with NaF: less with HRT and NaF [abstract]. Osteoporos Int 1996;6 Suppl 1:263.

80. Gutteridge DH, Kent GN, Prince RL, Nicholson GC, Stewart GO, Jones CE, et al. Fluoride treatment of osteoporosis: cyclical non-blinded or continuous blinded studies?Osteoporos Int 1993;3 Suppl 1:215–17.

81. Gutteridge DH, Price RI, Prince RL, Jones CE,Stewart GO, Nicholson GC, et al. Fluoride ±estrogen in osteoporosis – spinal bone gain, lowerlimb bone loss and stress fractures. J Bone Miner Res 1993;8 Suppl 1:S341.

82. Begg CB. Publication bias. In: Cooper H, Hedges LV, editors. The handbook of researchsynthesis. New York: Russell Sage Foundation 1994. p.339–409.


References

116

83. Windeler J, Lange S. Events per person year – a dubious concept. BMJ 1995;310:454–6.

84. Ross PD, Davis JW, Epstein RS, Wasnich RD. Pre-existing fractures and bone mass predict vertebralfracture incidence in women. Ann Intern Med1991;114:919–23.

85. Black DM, Arden NK, Palermo L, Pearson J,Cummings SR. Prevalent vertebral deformitiespredict hip fractures and new vertebral deformitiesbut not wrist fractures. J Bone Miner Res1999;14:821–8.

86. Meunier PJ. Evidence-based medicine andosteoporosis: a comparison of fracture riskreduction data from osteoporosis randomisedclinical trials. Int J Clin Pract 1999;53:122–9.

87. Cucherat M. EasyMA 97b. Software for meta-analysis of clinical trials. Lyon, France:1992–97. Available from: http://www.spc.univ-lyon1.fr/~mcu/easyma/

88. Detsky AS, Naylor CD, O’Rourke K, McGeer AJ,l’Abbe KA. Incorporating variations in the quality of individual randomized trials into meta-analysis. J Clin Epidemiol 1992;45:255–65.

89. Gillespie WJ, Henry DA, O’Connell DL, Robertson J. Vitamin D and vitamin D analoguesfor preventing fractures associated with involutionaland post-menopausal osteoporosis [Cochranereview]. In: The Cochrane Library. Issue 1. Oxford: Update Software: 1999.

90. Prendiville W, Elbourne D, Chalmers I. The effects of routine oxytocic administration in themanagement of the third stage of labour: anoverview of the evidence from controlled trials. Br J Obstet Gynaecol 1988;95:3–16.

91. Berlin JA. Does blinding of readers affect the results of meta-analyses? Lancet 1997;350:185–6.

92. Clark HD, Wells GA, Huet C, McAlister FA, Salmi LR, Fergusson D, et al. Assessing the quality of randomized trials: reliability of the Jadad scale.Control Clin Trials 1999;20:448–52.

93. Adami S, Passeri M, Ortolani S, Broggini M,Carratelli L, Caruso I, et al. Effects of oralalendronate and intranasal salmon calcitonin onbone mass and biochemical markers of boneturnover in postmenopausal women withosteoporosis. Bone 1995;17:383–90.

94. Bone HG, Downs RW Jr, Tucci JR, Harris ST,Weinstein R, Licata AA, et al. Dose-responserelationships for alendronate treatment inosteoporotic elderly women. J Clin Endocrinol Metab 1997;82:265–74.

95. Carfora E, Sergio F, Bellini P, Sergio C, Falco D,Zarcone R. Effect of treatment of postmenopausalosteoporosis with continuous daily oral alendronateand the incidence of fractures. Gazzetta Med ItalArch Sci Med 1998;157(4):105–9.

96. Chesnut CH 3rd, McClung MR, Ensrud KE, Bell NH, Genant HK, Harris ST, et al. Alendronatetreatment of the postmenopausal osteoporoticwoman: effect of multiple dosages on bone massand bone remodeling. Am J Med 1995;99:144–52.

97. Clemmesen B, Ravn P, Zegels B, Taquet AN,Christiansen C, Reginster JY. A 2-year phase II studywith 1-year of follow-up of risedronate (NE-58095)in postmenopausal osteoporosis. Osteoporos Int1997;7:488–95.

98. Black DM, Cummings SR, Karpf DB, Cauley JA,Thompson DE, Nevitt MC, et al. Randomised trialof effect of alendronate on risk of fracture inwomen with existing vertebral fractures. Lancet 1996;348:1535–41.

99. Cummings SR, Black DM, Thompson DE,Applegate WB, Barrett-Connor E, Musliner TA, et al. Effect of alendronate on risk of fracture inwomen with low bone density but without vertebralfractures: results from the Fracture InterventionTrial. JAMA 1998;280:2077–82.

100. Harris ST, Watts NB, Genant HK, McKeever CD,Hangartner T, Keller M, et al. Effects of risedronatetreatment on vertebral and nonvertebral fractures inwomen with postmenopausal osteoporosis: arandomized controlled trial. JAMA 1999;282: 1344–52.

101. Liberman UA, Weiss SR, Broll J, Minne HW, Quan H, Bell NH, et al. Effect of oral alendronateon bone mineral density and the incidence offractures in postmenopausal osteoporosis. N Engl J Med 1995;333:1437–43.

102. Lindsay R, Cosman F, Lobo RA, Walsh BW, Harris ST, Reagan JE, et al. Addition of alendronate toongoing hormone replacement therapy in thetreatment of osteoporosis: a randomized, controlledclinical trial. J Clin Endocrinol Metab 1999;84: 3076–81.

103. Montessori ML, Scheele WH, Netelenbos JC,Kerkhoff JF, Bakker K. The use of etidronate andcalcium versus calcium alone in the treatment ofpostmenopausal osteopenia: results of three yearsof treatment. Osteoporos Int 1997;7:52–8.

104. Pacifici R, McMurtry C, Vered I, Rupich R, Avioli LV. Coherence therapy does not prevent axial bone loss in osteoporotic women: a prelim-inary comparative study. J Clin Endocrinol Metab1988;66:747–53.

105. Pols HA, Felsenberg D, Hanley DA, Stepan J,Munoz-Torres M, Wilkin TJ, et al. Multinational,placebo-controlled, randomized trial of the effectsof alendronate on bone density and fracture risk in postmenopausal women with low bone mass:results of the FOSIT study. Osteoporos Int 1999;9:461–8.

106. Reginster J, Minne HW, Sorensen OH, Hooper M,Roux C, Brandi ML, et al. Randomized trial of theeffects of risedronate on vertebral fractures inwomen with established postmenopausalosteoporosis. Osteoporos Int 2000;11:83–91.


117

107. Reid IR, Wattie DJ, Evans MC, Gamble GD,Stapleton JP, Cornish J. Continuous therapy with pamidronate, a potent bisphosphonate, inpostmenopausal osteoporosis. J Clin EndocrinolMetab 1994;79:1595–9.

108. Storm T, Thamsborg G, Steiniche T, Genant HK,Sorensen OH. Effect of intermittent cyclicaletidronate therapy on bone mass and fracture rate in women with postmenopausal osteoporosis. N Engl J Med 1990;322:1265–71.

109. Watts NB, Harris ST, Genant HK, Wasnich RD,Miller PD, Jackson RD, et al. Intermittent cyclicaletidronate treatment of postmenopausalosteoporosis. N Engl J Med 1990;323:73–9.

110. Wimalawansa SJ. A four-year randomized controlledtrial of hormone replacement and bisphosphonate,alone or in combination, in women with post-menopausal osteoporosis. Am J Med1998;104:219–26.

111. Harris ST, Watts NB, Jackson RD, Genant HK,Wasnich RD, Ross P, et al. Four-year study ofintermittent cyclic etidronate treatment ofpostmenopausal osteoporosis: three years ofblinded therapy followed by one year of opentherapy. Am J Med 1993;95:557–67.

112. McCloskey EV, Spector TD, Eyres KS, Fern ED,O’Rourke N, Vasikaran S, et al. The assessment ofvertebral deformity: a method for use in populationstudies and clinical trials. Osteoporos Int 1993;3:138–47.

113. de Groen PC, Lubbe DF, Hirsch LJ, Daifotis A,Stephenson W, Freedholm D, et al. Esophagitisassociated with the use of alendronate. N Engl J Med 1996;335:1016–21.

114. Miller PD, Watts NB, Licata AA, Harris ST, Genant HK, Wasnich RD, et al. Cyclical etidronatein the treatment of postmenopausal osteoporosis:efficacy and safety after seven years of treatment.Am J Med 1997;103:468–76.

115. Thiébaud D, Burckhardt P, Melchior J, Eckert P,Jacquet A, Schnyder P, et al. Two years’ effectivenessof intravenous pamidronate (APD) versus oralfluoride for osteoporosis occurring in thepostmenopause. Osteoporos Int 1994;4:76–83.

116. Ettinger B, Pressman A, Schein J, Silver P, Chan JConnolly N. Survey of women taking alendronate:prevalence of non-compliance with instructions anddiscontinuation. Osteoporos Int 1997;7:46.

117. Aloia JF, Vaswani A, Yeh JK, Ellis K, Yasumura S,Cohn SH. Calcitriol in the treatment of post-menopausal osteoporosis. Am J Med 1988;84:401–8.

118. Gallagher JC, Riggs BL, Recker RR, Goldgar D. The effect of calcitriol on patients with post-menopausal osteoporosis with special reference to fracture frequency. Proc Soc Exp Biol Med1989;3:287–92.

119. Gallagher JC, Goldgar D. Treatment ofpostmenopausal osteoporosis with high doses ofsynthetic calcitriol. A randomized controlled study.Ann Intern Med 1990;113:649–55.

120. Orimo H, Shiraki M, Hayashi T, Nakamura T.Reduced occurrence of vertebral crush fractures in senile osteoporosis treated with 1 alpha (OH)-vitamin D3. Bone Miner 1987;3:47–52.

121. Orimo H, Shiraki M, Hayashi Y, Hoshino T, Onaya T, Miyazaki S, et al. Effects of 1-alpha-hydroxyvitamin D3 on lumbar bone mineral density and vertebral fractures in patients withpostmenopausal osteoporosis. Calcif Tissue Int1994;54:370–6.

122. Ott SM, Chesnut CH 3rd. Calcitriol treatment is not effective in postmenopausal osteoporosis [seecomments]. Ann Intern Med 1989;110:267–74.

123. Shiraki M, Kushida K, Yamazaki K, Nagai T, Inoue T, Orimo H. Effects of 2 years’ treatment ofosteoporosis with 1 alpha-hydroxy vitamin D3 onbone mineral density and incidence of fracture: aplacebo-controlled, double-blind prospective study.Endocrine J 1996;43:211–20.

124. Tilyard MW, Spears GF, Thomson J, Dovey S.Treatment of postmenopausal osteoporosis withcalcitriol or calcium. N Engl J Med 1992;326:357–62.

125. Levernieux J, Julien D, Caulin F. The effect ofcalcitonin on bone pain and acute resorptionrelated to recent osteoporotic crush fractures –result of a double-blind and an open study. In:Cecchettin M, Segre G, editors. Calciotropichormones and calcium metabolism. Amsterdam:Elsevier; 1986. p.171–8.

126. Lyritis GP, Tsakalakos N, Magiasis B, Karachalios T,Yiatzides A, Tsekoura M. Analgesic effect of salmoncalcitonin in osteoporotic vertebral fractures: adouble-blind placebo-controlled clinical study. Calcif Tissue Int 1991;49:369–72.

127. Lyritis GP, Paspati I, Karachalios T, Ioakimidis D,Skarantavos G, Lyritis PG. Pain relief from nasalsalmon calcitonin in osteoporotic vertebral crushfractures. A double-blind, placebo-controlledclinical study. Acta Orthop Scand Suppl1997;275:112–14.

128. Cristallini S, Pedetti M, Donatelli C, Gregorio F,Filipponi P. [The effect of different “coherent”-typetherapeutic plans on bone mineral density and onthe incidence of vertebral fractures in advancedosteoporosis. In Italian.] Recent Prog Med1993;84:336–45.

129. Ellerington MC, Hillard TC, Whitcroft SI, Marsh MS, Lees B, Banks LM, et al. Intranasalsalmon calcitonin for the prevention and treatmentof postmenopausal osteoporosis. Calcif Tissue Int1996;59:6–11.


References

118

130. Hizmetli S, Elden H, Kaptanoglu E, Nacitarhan V,Kocagil S. The effect of different doses of calcitoninon bone mineral density and fracture risk inpostmenopausal osteoporosis. Int J Clin Pract1998;52:453–5.

131. Hodsman AB, Fraher LJ, Watson PH, Ostbye T, Stitt LW, Adachi JD, et al. A randomized controlledtrial to compare the efficacy of cyclical parathyroidhormone versus cyclical parathyroid hormone andsequential calcitonin to improve bone mass inpostmenopausal women with osteoporosis. J ClinEndocrinol Metab 1997;82:620–8.

132. Pontiroli AE, Pajetta E, Calderara A, Alberetto M,Pozza G, Manganelli V, et al. Intranasal andintramuscular human calcitonin in femaleosteoporosis and in Paget’s disease of bones: a pilot study. J Endocrinol Invest 1991;14:47–51.

133. Stock JL, Avioli LV, Baylink DJ, Chesnut C, Genant HK, Maricic MJ, et al. Calcitonin-salmonnasal spray reduces the incidence of new vertebralfractures in post-menopausal women: three yearinterim results of the PROOF study. J Bone Miner Res 1997;12:S149.

134. Rico H, Hernandez ER, Revilla M, Gomez-Castresana F. Salmon calcitonin reduces vertebralfracture rate in postmenopausal crush fracturesyndrome. Bone Miner 1992;16:131–8.

135. Rico H, Revilla M, Hernandez ER, Villa LF, Alvarez de Buergo M. Total and regional bonemineral content and fracture rate in postmenopausalosteoporosis treated with salmon calcitonin: aprospective study. Calcif Tissue Int 1995;56:181–5.

136. Ringe JD, Welzel D. Salmon calcitonin in thetherapy of corticoid-induced osteoporosis. Eur J Clin Pharmacol 1987;33:35–9.

137. Ringe JD. [Treatment of primary osteoporosis with calcium and salmon calcitonin. In German.]Dtsch Med Wochensch 1990;115:1176–82.

138. Kanis JA, McCloskey EV. Effect of calcitonin onvertebral and other fractures. QJM 1999;92:143–9.

139. Abellan Perez M, Bayina Garcia FJ, Calabozo M,Carpintero Benitez P, Figueroa Pedrosa M,Fernandez Crisostomo C, et al. [Multicentrecomparative study of synthetic salmon calcitoninadministered nasally in the treatment of establishedpostmenopausal osteoporosis. In Spanish.] An MedInterna 1995;12:12–16.

140. Grazioli I, Strumia E. Drug surveillance concerningcalcitonin: preliminary findings on efficacy andtolerability. Clin Ter 1988;127:289–95.

141. Osteoporosis: clinical guidelines for prevention andtreatment. London: Royal College of Physicians ofLondon; 1999.

142. Recker RR, Hinders S, Davies KM, Heaney RP,Stegman MR, Lappe JM, et al. Correcting calciumnutritional deficiency prevents spine fractures inelderly women. J Bone Miner Res 1996;11:1961–6.

143. Lufkin EG, Wahner HW, O’Fallon WM, Hodgson SF, Kptpwocz MA, Lane AW, et al.Treatment of postmenopausal osteoporosis withtransdermal estrogen. Ann Intern Med 1992;117:1–9.

144. Zarcone R, Carfora E, Sergio F, Bellini P, Longo M,Tartaglia E, et al. [Oestrogen therapy in women withpostmenopausal osteoporosis. In Italian.] MinervaGinecol 1997;49:355–9.

145. Barrett-Connor E. Fortnightly review – hormonereplacement therapy. BMJ 1998;317:457–61.

146. Grodstein F, Newcomb PA, Stampfer MJ.Postmenopausal hormone therapy and the risk of colorectal cancer: a review and meta-analysis. Am J Med 1999;106:574–82.

147. Sherwin BB. Hormones, mood, and cognitivefunctioning in postmenopausal women. ObstetGynecol 1996;87:20–6S.

148. Yaffe K, Sawaya G, Lieberburg I, Grady D. Estrogen therapy in postmenopausal women: effects on cognitive function and dementia. JAMA 1998;279:688–95.

149. Collaborative Group on Hormonal Factors in BreastCancer. Breast cancer and hormone replacementtherapy: collaborative reanalysis of data from 51epidemiological studies of 52,705 women withbreast cancer and 108,411 women without breastcancer. Lancet 1997;350:1047–59.

150. Grady D, Wenger NK, Herrington D, Khan S,Furberg C, Hunninghake D, et al. Postmenopausalhormone therapy increases risk for venousthromboembolic disease. Ann Intern Med2000;132:689–96.

151. Grady D, Rubin SM, Pettiti DB. Hormone therapy to prevent disease and prolong life in postmenopausal women. Ann Intern Med1992;117:1016–37.

152. Grady D, Gebretsadik T, Kerlikowske K, Ernster V,Pettiti D. Hormone replacement therapy andendometrial cancer risk: a meta-analysis. ObstetGynecol 1995;85:304–13.

153. Hulley S, Grady D, Bush T, Furberg C, Herrington D, Riggs BL, et al. Randomized trial of estrogen plus progestin for secondary preventionof coronary heart disease in postmenopausalwomen. JAMA 1998;280:605–13.

154. Wallace WA, Price VH, Elliot CA, MacPherson MB,Scott BW. Hormone replacement therapy accept-ability to Nottingham post-menopausal women with a risk factor for osteoporosis. J R Soc Med1990;83:699–701.

155. Torgerson DJ, Donaldson C, Russell IT, Reid DM.Hormone replacement therapy: compliance andcost after screening for osteoporosis. Eur J ObstetGynecol Reprod Biol 1995;59:57–60.


119

156. Ryan PJ, Harrison R, Blake GM, Fogelman I.Compliance with hormone replacement therapy(HRT) after screening for post menopausalosteoporosis. Br J Obstet Gynaecol 1992;99:325–8.

157. Maugeri D, Panebianco P, Russo MS, Motta M,Tropea S, Motta L, et al. Ipriflavone-treatment ofsenile osteoporosis: results of a multicenter, double-blind clinical trial of 2 years. Arch Gerontol Geriatr1994;19:253–63.

158. Passeri M, Biondi M, Costi D, Bufalino L,Castiglione GN, di Peppe C, et al. Effect ofipriflavone on bone mass in elderly osteoporoticwomen. Bone Miner 1992;19 Suppl 1:S57–62.

159. Passeri M, Biondi M, Costi D, Dall’Aglio E,Pedrazzoni M, Bufalino L, et al. Effects of 2-yeartherapy with ipriflavone in elderly women withestablished osteoporosis. Ital J Miner Electrolyte Metab 1995;9:137–44.

160. Agnusdei D, Bufalino L. Efficacy of ipriflavone in established osteoporosis and long-term safety.Calcif Tissue Int 1997;61 Suppl 1:S23–7.

161. Chesnut CH 3rd, Ivey JL, Gruber HE, Matthews M,Nelp WB, Sisom K, et al. Stanozolol in postmenopausalosteoporosis: therapeutic efficacy and possiblemechanisms of action. Metabolism 1983;32:571–80.

162. Geusens P. Nandrolone decanoate: pharmacologicalproperties and therapeutic use in osteoporosis. Clin Rheumatol 1995;14 Suppl 3:32–9.

163. Geusens P, Dequeker J. Long-term effect ofnandrolone decanoate, 1-alpha-hydroxy vitamin D3or intermittent calcium infusion therapy on bonemineral content, bone remodeling and fracturerate in symptomatic osteoporosis: a double-blindcontrolled study. Bone Miner 1986;1:347–57.

164. Birkenhager JC, Erdtsieck RJ, Zeelenberg J, van Kuik C, van Veen LC, Birkenhager-Frenkel DH,et al. Can nandrolone add to the effect of hormonalreplacement therapy in postmenopausalosteoporosis? Bone Miner 1992;18:251–65.

165. Need AG, Horowitz M, Walker CJ, Chatterton BE,Chapman JC, Nordin BE. Crossover study of fat-corrected forearm mineral content duringnandrolone decanoate therapy for osteoporosis.Bone 1989;10:3–6.

166. Lyritis GP, Androulakis C, Magiasis B, Charalambaki Z,Tsakalakos N. Effect of nandrolone decanoate and 1-alpha-hydroxy-calciferol on patients with vertebralosteoporotic collapse. A double-blind clinical trial.Bone Miner 1994;27:209–17.

167. Hansson T, Roos B. The effect of fluoride andcalcium on spinal bone mineral content: acontrolled, prospective (3 years) study. Calcif Tissue Int 1987;40:315–17.

168. Kleerekoper M, Peterson EL, Nelson DA, Phillips E,Schork MA, Tilley B, et al. A randomized trial ofsodium fluoride as a treatment for postmenopausalosteoporosis. Osteoporos Int 1991;1:155–61.

169. Meunier PJ, Sebert JL, Reginster JY, Briancon D,Appelboom T, Netter P, et al. Fluoride salts are nobetter at preventing new vertebral fractures thancalcium-vitamin D in postmenopausal osteoporosis:the FAVO study. Osteoporos Int 1998;8:4–12.

170. Pak CY, Sakhaee K, Adams-Huet B, Piziak V,Peterson RD, Poindexter JR. Treatment ofpostmenopausal osteoporosis with slow-releasesodium fluoride. Final report of a randomizedcontrolled trial. Ann Intern Med 1995;123:401–8.

171. Reginster JY, Meurmans L, Zegels B, Rovati LC,Minne HW, Giacovelli G, et al. The effect of sodiummonofluorophosphate plus calcium on vertebralfracture rate in postmenopausal women withmoderate osteoporosis. A randomized, controlledtrial. Ann Intern Med 1998;129:1–8.

172. Riggs BL, Hodgson SF, O’Fallon WM, Chao EW,Wahner HW, Muhs JM, et al. Effect of fluoridetreatment on the fracture rate in postmenopausalwomen with osteoporosis. N Engl J Med1990;322:802–9.

173. Ringe JD, Dorst A, Kipshoven C, Rovati LC, Setnikar I. Avoidance of vertebral fractures in men with idiopathic osteoporosis by a three yeartherapy with calcium and low-dose intermittentmonofluorophosphate. Osteoporos Int 1998;8:47–52.

174. Ringe JD, Kipshoven C, Coster A, Umbach R.Therapy of established postmenopausalosteoporosis with monofluorophosphate pluscalcium: dose-related effects on bone density andfracture rate. Osteoporos Int 1999;9:171–8.

175. Sebert JL, Richard P, Mennecier I, Bisset JP, Loeb G. Monofluorophosphate increases lumbarbone density in osteopenic patients: a double-masked randomized study. Osteoporos Int1995;5:108–14.

176. Mamelle N, Meunier PJ, Dusan R, Guillaume M,Martin JL, Gaucher A, et al. Risk-benefit ratio ofsodium fluoride treatment in primary vertebralosteoporosis. Lancet 1988;2:361–5.

177. Ettinger B, Black DM, Mitlak BH, Knickerbocker RK, Nickelsen T, Genant H, et al.Reduction of vertebral fracture risk in post-menopausal women with osteoporosis treated with raloxifene: results from a 3-year randomizedclinical trial. JAMA 1999;282:637–45.

178. Lufkin EG, Whitaker MD, Nickelsen T, Argueta R,Caplan RH, Knickerbocker RK, et al. Treatment of established postmenopausal osteoporosis withraloxifene: a randomized trial. J Bone Miner Res1998;13:1747–54.

179. Delmas PD, Mitlak BH, Christiansen C. Effects ofraloxifene in postmenopausal women. N Engl J Med1998;338:1313–14.


References

120

180. Walsh BW, Kuller LH, Wild RA, Paul S, Farmer M,Lawrence JB, et al. Effects of raloxifene on serumlipids and coagulation factors in healthy post-menopausal women. JAMA 1998;279:1445–51.

181. Delmas PD, Bjarnason NH, Mitlak BH, Ravoux AC,Shah AS, Huster WJ, et al. Effects of raloxifene on bone mineral density, serum cholesterol con-centrations, and uterine endometrium in post-menopausal women. N Engl J Med 1997;337:1641–7.

182. Cummings SR, Eckert S, Krueger KA, Grady D,Powles TJ, Cauley JA, et al. The effect of raloxifeneon risk of breast cancer in postmenopausal women:results from the MORE randomized trial. JAMA1999;281:2189–97.

183. Khovidhunkit W, Shoback DM. Clinical effects ofraloxifene hydrochloride in women. Ann Intern Med1999;130:431–9.

184. Jordan VC, Glusman JE, Eckert S, Lippman M,Powles T, Costa A, et al. Incident primary breastcancers are reduced by raloxifene: integrated datafrom multicenter, double-blind, randomised trialsin ~12,000 postmenopausal women. Proc Am SocClin Oncol 1998;17:122a.

185. Nickelsen T, Lufkin EG, Riggs BL, Cox DA, Crook TH. Raloxifene hydrochloride, a selectiveestrogen receptor modulator: safety assessment of effects on cognitive function and mood inpostmenopausal women. Psychoneuroendocrinology1999;24:115–28.

186. Schurch MA, Rizzoli R, Slosman D, Vadas L,Vergnaud P, Bonjour JP. Protein supplementsincrease serum insulin-like growth factor-I levelsand attenuate proximal femur bone loss in patientswith recent hip fracture. A randomized, double-blind, placebo-controlled trial. Ann Intern Med1998;128:801–9.

187. Ebrahim S, Thompson PW, Baskaran V, Evans K.Randomized placebo-controlled trial of briskwalking in the prevention of postmenopausalosteoporosis. Age Ageing 1997;26:253–60.

188. Hardman AE, Hudson A, Jones PRM, Norgan NG.Brisk walking and plasma high density lipoproteincholesterol concentration in previously sedentarywomen. BMJ 1989;299:1204–5.

189. Leon AS, Conrad J, Hunninghake DB, Serfass R.Effects of a vigorous walk programme on bodycomposition and carbodydrate and lipidmetabolism of obese young men. Am J Clin Nutr1979;32:1776–87.

190. Arthur RS, Piraino B, Candib D, Cooperstein L,Chen T, West C, et al. Effect of low-dose calcitrioland calcium therapy on bone histomorphometryand urinary calcium excretion in osteopenicwomen. Miner Electrolyte Metab 1990;16:385–90.

191. Falch JA, Odegaard OR, Finnanger AM, Matheson I. Postmenopausal osteoporosis: no effect of three years treatment with 1,25-dihydroxycholecalciferol. Acta Med Scand1987;221:199–204.

192. Lems WF, Jacobs JW, Bijlsma JW, van Veen GJ,Houben HH, Haanen HC, et al. Is the addition of sodium fluoride to cyclical etidronate beneficialin the treatment of corticosteroid inducedosteoporosis? Ann Rheum Dis 1997;56:357–63.

193. Pak CY, Sakhaee K, Zerwekh JE, Parcel C, Peterson R, Johnson K. Safe and effective treatment of osteoporosis with intermittent slowrelease sodium fluoride: augmentation of vertebralbone mass and inhibition of fractures. J ClinEndocrinol Metab 1989;68:150–9.

194. Shiota E. Evaluation of the drug therapy forestablished osteoporosis by dual-energy x-rayabsorptiometry. Fukuoka Igaku Zasshi 1998;89:172–8.

195. Shiraki M, Kushid K, Fukunaga M, Kishimoto H,Taga M, Nakamura T, et al. A double-maskedmulticenter comparative study between alendronateand alfacalcidol in Japanese patients withosteoporosis. Osteoporos Int 1999;10:183–92.

196. Black DM, Palermo L, Nevitt MC, Genant HK,Christensen L, Cummings SR. Defining incidentvertebral deformity: a prospective comparison ofseveral approaches. J Bone Miner Res1999;14:90–101.

197. Michaelsson K, Baron JA, Farahmand BY, Johnell O,Magnusson C, Persson PG, et al. Hormone replace-ment therapy and risk of hip fracture: populationbased case–control study. BMJ 1998;316:1858–63.

198. Dawson-Hughes B, Harris SS, Krall EA, Dallal GE.Effect of withdrawal of calcium and vitamin Dsupplements on bone mass in elderly men andwomen. Am J Clin Nutr 2000;72:745–50.

199. Devine A, Dick IM, Heal SJ, Criddle RA, Prince RL.A 4-year follow-up study of the effects of calciumsupplementation on bone density in elderlypostmenopausal women. Osteoporos Int 1997;7:23–8.

200. Pouilles JM, Tremollieres F, Ribot C. Prevention ofpost-menopausal bone loss with 1-alpha-hydroxyvitamin D3. A three-year prospective study. ClinRheumatol 1992;11:492–7.

201. Lindsay R, Hart DM, Maclean A, Clark AC, KraszewskiA, Garwood J. Bone response to termination ofoestrogen treatment. Lancet 1978;i:1325–7.

202. Stevenson JC, Kanis JA, Christiansen C. Bone-density measurement. Lancet 1992;339:370–1.

203. Weiss NS, Ure CL, Ballard JH, Williams AR, Daling JR. Decreased risk of fractures of the hipand lower forearm with postmenopausal use ofestrogen. N Engl J Med 1980;303:1195–8.


121

204. Kristensen B, Ejlertsen B, Mouridsen HT, Andersen KW, Lauritzen JB. Femoral fractures in postmenopausal breast cancer patients treatedwith adjuvant tamoxifen. Breast Cancer Res Treat1996;39:321–6.

205. Felson DT, Zhang Y, Hannan MT, Kiel DP, Wilson PW, Anderson JJ. The effect of post-menopausal estrogen therapy on bone density inelderly women. N Engl J Med 1993;329:1141–6.

206. Keil DP, Felson DT, Anderson JJ. Hip fracture andthe use of estrogens in postmenopausal women. N Engl J Med 1987;317:1169–74.

207. Haji-Fuleihan G, Luckery M, Bauer R, Wasnich R,Krupa D, Reda C, et al. Effects of discontinuation of alendronate in early postmenopausal women. J Bone Miner Res 1997;12:S143.

208. Khan SA, Kanis JA, Vasikaran S, Kline WF,Matuszewski BK, McCloskey EV, et al. Eliminationand biochemical responses to intravenousalendronate in postmenopausal osteoporosis. J Bone Miner Res 1997;12:1700–7.

209. Stock JL, Bell NH, Chesnut CH 3rd, Ensrud KE,Genant HK, Harris ST, et al. Increments in bonemineral density of the lumbar spine and hip andsuppression of bone turnover are maintained afterdiscontinuation of alendronate in postmenopausalwomen. Am J Med 1997;103:291–7.

210. Landman JO, Hamdy NA, Pauwels EK, Papapoulos SE. Skeletal metabolism in patients with osteoporosis after discontinuation of long-term treatment with oral pamidronate. J ClinEndocrinol Metab 1995;80:3465–8.

211. Wasnich R, Davis J, Workman P, et al. Effects ofalendronate discontinuation on BMD in earlypostmenopausal women. J Bone Miner Res1998;23:S402.

212. Delmas PD, Balena R, Confravreux E, Hardouin C,Hardy P, Bremond A. Bisphosphonate risedronateprevents bone loss in women with artificialmenopause due to chemotherapy of breast cancer:a double-blind, placebo-controlled study. J ClinOncol 1997;15:955–62.

213. Talbot JR, Fischer MM, Farley SM, Libanati C, Farley J, Tabuenca A, et al. The increase in spinalbone density that occurs in response to fluoridetherapy for osteoporosis is not maintained after the therapy is discontinued. Osteoporos Int1996;6:442–47.

214. Johnell O, Steinbeck M, Rosen M, Gullberg B, Kanis JA. Therapeutic strategies in the preventionof hip fracture with drugs affecting bonemetabolism. Bone 1993;14:S85–7.

215. Johnell O, Oden A, Kanis JA, Caulin F. Prevention ofosteoporotic fractures. Calcif Tissue Int 1997;61:500.

216. Torgerson DJ, Reid DM. The economics ofosteoporosis and its prevention. A review.PharmacoEconomics 1997;11:126–38.

217. Daley E, Vessey MP, Barlow D, Gray A, McPherson K, Roche M. Hormone replacementtherapy in a risk-benefit perspective. In: Berg G,Hammar M, editors. The modern management ofthe menopause: a perspective for the 21st century.Proceedings Second International Congress on the Menopause, Stockholm, June 20–24, 1994.London: Parthenon; 1994. p.473–97.

218. Roche M, Vessey M. Hormone replacement therapy in the menopause – risks, benefits andcosts. In: Drife JO, Studd JWW, editors. HRT and osteoporosis. 1990. p.363–72.

219. Weinstein MC, Schiff I. Cost-effectiveness ofhormone replacement therapy in the menopause.Obstet Gynecol Surv 1983;38:445–55.

220. Cheung AP, Wren BG. A cost-effectiveness analysisof hormone replacement therapy in themenopause. Med J Aust 1992;156:312–16.

221. McClung MR, Geusens P, Miller P, Zippel H, Bensen WG, Roux C, et al. Effect of risedronate on the risk of hip fracture in elderly women. N Engl J Med 2001;344:333–40.

222. Chapuy MC, Arlot ME, Duboeuf F, Brun J, Crouzet B, Arnaud S, et al. Vitamin D3 and calcium to prevent hip fractures in elderly women.N Engl J Med 1992;327:1637–42.

223. Heikinheimo RJ, Inkovaara JA, Harju EJ, Haavisto MV, Kaarela RH, Kataja JH, et al. Annual injection of vitamin D and fractures of aged bones. Calcif Tissue Int 1992;51:105–10.

224. Lips P, Graafmans WC, Ooms ME, Bezemer PD,Bouter LM. Vitamin D supplementation andfracture incidence in elderly persons. Arandomized, placebo-controlled clinical trial. Ann Intern Med 1996;124:400–6.

225. Koumulainen MH, Kroger H, Tuppurainen MT,Heikkinen AM, Alhava E, Honkanen R, et al. HRTand vit D in prevention of non-vertebral fractures inpostmenopausal women: a 5 year randomized trial.Maturitas 1998;31:45–54.

226. Naessen T, Persson I, Adami HO, Bergstrom R,Bergkvist L. Hormone replacement therapy and therisk for first hip fracture. A prospective, population-based cohort study. Ann Intern Med 1990;113:95–103.

227. Cauley JA, Seeley DG, Ensrud K, Ettinger B, Black D, Cummings SR. Estrogen replacementtherapy and fractures in older women. Ann InternMed 1995;122:9–16.

228. Porter M, Penney GC, Russell D, Russell E,Templeton A. A population based survey ofwomen’s experience of the menopause. Br J Obstet Gynaecol 1996;103:1025–8.

229. Rymer J, Morris EP. Menopausal symptoms. BMJ 2000;321:1516–19.


References

122

230. Barrett-Connor E, Grady D. Hormone replacementtherapy, heart disease, and other considerations.Annu Rev Public Health 1998;19:55–72.

231. Posthuma WFM, Westendorp RGJ, Vandenbroucke JP.Cardioprotective effect of hormone replacementtherapy in postmenopausal women: is the evidencebiased? BMJ 1994;308:1268–9.

232. Hemminki E, McPherson K. Impact of post-menopausal hormone therapy on cardiovascularevents and cancer: pooled data from clinical trials.BMJ 1997;315:149–53.

233. Grodstein F, Stampfer MJ, Colditz GA, Willett WC,Manson JE, Joffe M, et al. Postmenopausal hormonetherapy and mortality. N Engl J Med1997;336:1769–75.

234. Stampfer MJ, Colditz GA, Willett WC, Manson JE,Rosner B, Speizer FE, et al. Postmenopausalestrogen therapy and cardiovascular disease. N Engl J Med 1991;325:756–62.

235. Vessey MP. Exogenous oestrogens and the risk of breast cancer. Proc R Coll Physicians Edinb1999;29:133–6.

236. Cauley JA, Seeley DG, Browner WS, Ensrud K,Kuller LH, Lipschutz RC, et al. Estrogen replace-ment therapy and mortality among older women.The study of osteoporotic fractures. Arch Intern Med1997;157:2181–7.

237. Perez Gutthann S, Garcia Rodrigues LA,Castellsague J, Duque Oliart A. Hormonereplacement therapy and risk of venousthromboembolism: population based case-controlstudy. BMJ 1997;314:796–800.

238. Daly E, Vessey MP, Hawkins MM, Carson JL, Gough P, Marsh S. Risk of venous thrombo-embolism in users of hormone replacementtherapy. Lancet 1996;348:977–80.

239. Jick H, Derby LE, Wald Myers M, Vasilakis C,Newton KM. Risk of hospital admission foridiopathic venous thromboembolism among users of postmenopausal oestrogens. Lancet1996;348:981–3.

240. Kanis JA, Beneton MNC, Gennari C, et al. Effects of anabolic steroids on cortical bone and fracture.In: Christiansen C, Riis B, editors. Osteoporosis1993. Proceedings 4th International Conference on Osteoporosis. Copenhagen, 1993. p.308–10.

241. Haguenauer D, Welch V, Shea B, Tugwell P, Adachi JD, Wells G. Fluoride for the treatment ofpostmenopausal osteoporotic fractures: a meta-analysis. Osteoporos Int 2000;11:728–38.

242. Wasnich R, Davis J, Ross P, Vogel J. Effect of thiazideon rates of bone mineral loss: a longitudinal study.BMJ 1990;301:1303–5.

243. Ray WA, Griffin MR, Downey W, Melton LJ 3rd.Long-term use of thiazide diuretics and risk of hipfracture. Lancet 1989;i:687–90.

244. Felson DT, Sloutskis D, Anderson JJ, Anthony JM,Kiel DP. Thiazide diuretics and the risk of hipfracture. Results from the Framingham study. JAMA 1991;265:370–3.

245. LaCroix AZ, Wienpahl J, White LR, Wallace RB,Scherr PA, George LK, et al. Thiazide diureticagents and the incidence of hip fracture. N Engl J Med 1990;322:286–90.

246. Jones G, Nguyen T, Sambrook PN, Eisman JA.Thiazide diuretics and fractures: can meta-analysishelp? J Bone Miner Res 1995;10:106–11.

247. Law MR, Wald NJ, Thompson SG. By how muchand how quickly does reduction in serumcholesterol concentration lower risk of ischaemicheart disease? BMJ 1994;308:367–72.

248. Sanders KM, Pasco JA, Ugoni AM, Nicholson GC,Seeman E, Martin TJ, et al. The exclusion of hightrauma fractures may underestimate the prevalenceof bone fragility fractures in the community: theGeelong osteoporosis study. J Bone Miner Res1998;13:1337–42.

249. Seeley DG, Browner WS, Nevitt MC, Genant HK,Scott JC, Cummings SR. Which fractures areassociated with low appendicular bone mass inelderly women? Ann Intern Med 1991;115:837–42.

250. Melton LJ, Atkinson EJ, Cooper C, O’Fallon WM,Riggs BL. Vertebral fractures predict subsequentfractures. Osteoporos Int 1999;10:214–21.

251. Kotowicz MA, Melton LJ 3rd, Cooper C, Atkinson EJ, O’Fallon WM, Riggs BL. Risk of hip fracture in women with vertebral fracture. J Bone Miner Res 1994;9:599–605.

252. Lippuner K, von Overbeck J, Perrelet R, Bosshard H, Jaeger P. Incidence and direct medical costs of hospitalizations due to osteo-porotic fractures in Switzerland. Osteoporos Int1997;7:414–25.

253. Phillips S, Fox N, Jacobs J, Wright W. The directmedical costs of osteoporosis for American womenaged 45 and older, 1986. Bone 1988;9:271–9.

254. Melton LJ 3rd, Thamer M, Ray NF, Chan JK,Chesnut CH 3rd, Einhorn TA, et al. Fracturesattributable to osteoporosis: report from theNational Osteoporosis Foundation. J Bone Miner Res 1997;12:16–23.

255. Donaldson LJ, Cook A, Thomson R. Incidence offractures in a geographically defined population. J Epidemiol Community Health 1990;44:241–5.

256. Johansen A, Evans RJ, Stone MD, Richmond PW, Lo SV, Woodhouse KW. Fracture incidence inEngland and Wales: a study based on thepopulation of Cardiff. Injury 1997;28:655–60.

257. Falch J, Kaastad TS, Bohler G, Espeland J,Sundsvold OJ. Secular increase and geographicdifferences in hip fracture incidence in Norway.Bone 1993;14:643–5.


123

258. Kanis JA, McCloskey EV. Epidemiology of vertebralosteoporosis. Bone 1992;13 Suppl 2:S1–10.

259. Donnelly S, Spector TD, Reaney L, Doyle DV, Kanis JA, McCloskey EV. Incidence of vertebralfracture in early postmenopausal women in theUnited Kingdom: the Chingford study. Osteoporos Int; in press.

260. Palvanen M, Kannus P, Niemi S, Parkkari J. Seculartrends in the osteoporotic fractures of the distalhumerus in elderly women. Eur J Epidemiol1998;14:159–64.

261. Frost ML, Blake GM, Fogelman I. Quantitativeultrasound and bone mineral density are equallystrongly associated with risk factors for osteoporosis.J Bone Miner Res 2001;16:406–14.

262. Kanis JA, Johnell O, Oden A, Jonsson B, de Laet C,Dawson A. Risk of hip fracture according to theWorld Health Organization criteria for osteopeniaand osteoporosis. Bone 2000;27:585–90.

263. Klotzbuecher CM, Ross PD, Landsman PB, Abbott TA, Berger M. Patients with prior fractureshave an increased risk of future fractures: asummary of the literature and statistical synthesis. J Bone Miner Res 2000;15:721–7.

264. Zamet JS, Darbar UR, Griffiths GS, Bulman JS,Bragger U, Burgin W, et al. Particulate bioglass as agrafting material in the treatment of periodontalintrabony defects. J Clin Periodontol 1997;24:410–18.

265. Johnell O, Oden A, Caulin F, Kanis JA. Acute andlong-term increase in fracture risk after hospital-isation for vertebral fracture. Osteoporos Int2001;12:207–14.

266. Parker MJ, Anand JK. What is the true mortality ofhip fracture? Public Health 1991;105:443–6.

267. Meyer HE, Tverdal A, Falch JA, Pedersen JI. Factorsassociated with mortality after hip fracture.Osteoporos Int 2000;11:228–32.

268. Poor G, Atkinson EJ, O’Fallon WM, Melton LJ.Determinants of reduced survival following hipfractures in men. Clin Orthop 1995;319:260–5.

269. Todd CJ, Freeman C, Camilleri-Ferrante C, Laxton C, Muriel P, Palmer C, et al. Anglian audit of hip fracture 2. Cambridge: CambridgeHealth Services Research Group, University of Cambridge; 1999.

270. Forsen L, Sogaard AJ, Meyer HE, Edna T, Kopjar B.Survival after hip fracture: short and long-termexcess mortality according to age and gender.Osteoporos Int 1999;10:73–8.

271. Laxton C, Freeman C, Todd C, Payne BV, Camilleri-Ferrante C. Morbidity at 3 months afterhip fracture: data from the East Anglian audit.Health Trends 1997;29:55–60.

272. Kado DM, Browner WS, Palermo L, Nevitt MC,Genant HK, Cummings SR. Vertebral fractures and mortality in older women: a prospective study.Arch Intern Med 1999;159:1215–20.

273. Center JR, Nguyen TV, Schneider D, Sambrook PN,Eisman JA. Mortality after all major types ofosteoporotic fracture in men and women: anobservational study. Lancet 1999;353:878–82.

274. Cooper C, Atkinson EJ, Jacobsen SJ, O’Fallon WM,Melton LJ 3rd. Population-based study of survivalafter osteoporotic fractures. Am J Epidemiol1993;137:1001–5.

275. Browner WS, Seeley DG, Vogt TM, Cummings SR.Non-trauma mortality in elderly women with lowbone mineral density. Lancet 1991;338:355–8.

276. Johansson C, Black D, Johnell O, Oden A,Mellstrom D. Bone mineral density is a predictor of survival. Calcif Tissue Int 1998;63:190–6.

277. Cauley JA, Thompson DE, Ensrud KC, Scott JC,Black D. Risk of mortality following clinicalfractures. Osteoporos Int 2000;11:556–61.

278. Office of National Statistics. Mortality Series. DH2 no 25. London: Stationery Office; 1998.

279. Office of National Statistics. Cancer survival trendsin England and Wales 1971–1995. Deprivation andNHS region. London: Stationery Office; 2001.

280. Cauley JA, Lucas FL, Kuller LH, Vogt MT, Browner WS, Cummings SR. Bone mineral densityand risk of breast cancer in older women: the studyof osteoporotic fractures. JAMA 1996;276:1404–8.

281. Zhang Y, Kiel DP, Kreger BE, Cupples LA, Ellison RC, Dorgan JF, et al. Bone mass and the risk of breast cancer among postmenopausalwomen. N Engl J Med 1997;336:611–17.

282. Volmink JA, Newton JN, Hicks NR, Sleight P, Fowler GH, Neil HAW. Coronary event and casefatality rates in an English population: results of the Oxford myocardial incidence study. Heart 1998;80:40–4.

283. Expectation of life: United Kingdom females –1999. London: Government Actuary Office; 1999.

284. Sculpher M, Torgerson D, Goeree R, O’Brien B. A critical structured review of economic evaluationsof interventions for the prevention and treatmentof osteoporosis. Discussion Paper 169. York: Uni-versity of York Centre for Health Economics; 1999.

285. Oleksik A, Lips P, Dawson A, Minshall ME, Shen W,Cooper C, et al. Health-related quality of life inpostmenopausal women with low BMD with orwithout prevalent vertebral fracture. J Bone MinerRes 2000;15:1384–92.

286. Brazier JE, Kohler B, Walters S. A prospective studyof the health related quality of life impact of hipfracture. Sheffield: SCHaRR, University of Sheffield; 2000.

287. Hutton J, Brown R, Borowitz M, Abrams K,Rothman M, Shakespeare A. A new decision modelfor cost–utility comparisons of chemotherapy inrecurrent metastatic breast cancer.PharmacoEconomics 1996;9:8–22.


References

124

288. Gabriel SE, Kneeland TS, Melton LJ 3rd, Moncur MM, Ettinger B, Tosteson AN. Health-related quality of life in economic evaluations for osteoporosis: whose values should we use? Med Decis Making 1999;19:141–8.

289. Gold DT. The clinical impact of vertebral fractures:quality of life in women with osteoporosis. Bone1996;18:185–9S.

290. Salkeld G, Cameron ID, Cumming RG, Easter S,Seymour J, Kurrie SE, et al. Quality of life related to fear of falling and hip fracture in older women: a time trade-off study. BMJ 2000;320:341–6.

291. Dolan P. Modelling valuation for Euroqol healthstates. Med Care 1997;35:351–63.

292. Fitzpatrick R, Zeibland S, Jenkinson C, Mowat A. A comparison of the sensitivity to change of severalhealth status measures in rheumatoid arthritis. J Rheumatol 1993;20:429–36.

293. Brazier J, Deverill M, Green C. A review of the useof health status measures in economic evaluation. J Health Serv Res Policy 1999;4:174–84.

294. Fryback DG, Dasbach EJ, Klein R, Klein BE, Dorn N, Peterson K, et al. The Beaver Dam healthoutcomes study: initial catalog of health-statequality factors. Med Decis Making 1993;13:89–102.

295. Roberge R, Berthelot J, Cranswick K. Adjusting lifeexpectancy to account for disability in the popu-lation: a comparison of three techniques. SocIndicators Res 1999;48:217–43.

296. Kind P, Dolan P, Gudex C, Williams A. Variations inpopulation health status: results from a UK nationalquestionnaire survey. BMJ 1998;316:736–41.

297. Dolan P, Torgerson D, Kakarlapudi TK. Health-related quality of life of Colles’ fracture patients.Osteoporos Int 1999;9:196–9.

298. Gold MR, Siegel JE, Russell LB. Cost-effectiveness inhealth and medicine. Oxford: Oxford UniversityPress; 1996.

299. Tengs TO, Wallace A. One thousand health related quality of life estimates. Med Care2000;38:583--637.

300. NHS Centre for Reviews and Dissemination.Available from: http://agatha.york.ac.uk/nf2.htm

301. Beech R, Withey C, Morris R. Understandingvariations in lengths of stay between hospitals forfractured neck of femur patients and the potentialconsequences of reduced stay targets. J Public HealthMed 1995;17:77–84.

302. Best L, Milne R. Bisphosphonates (alendronate andetidronate) in the management of osteoporosis.DEC report no. 79. Southampton: Wessex Institutefor Health Research and Development; 1998.

303. Drummond MF, Heyse J, Cook J, McGuire A.Selection of end points in economic evaluations ofcoronary-heart-disease interventions. Med DecisMaking 1993;13:184–90.

304. Fox HJ, Hughes SJ, Pooler J, Prothero D, Bannister GC. Length of hospital stay and outcomeafter femoral neck fracture: a prospective studycomparing the performance of two hospitals. Injury 1993;24:464–6.

305. Francis RM, Anderson FH, Torgerson DJ. Acomparison of the effectiveness and cost oftreatment for vertebral fractures in women. Br J Rheumatol 1995;34:1167–71.

306. French FH, Torgerson DJ, Porter RW. Cost analysisof fracture of the neck of femur. Age Ageing1995;24:185–9.

307. Hollingworth W, Todd CJ, Parker MJ. The cost oftreating hip fractures in the twenty-first century. J Public Health Med 1995;17:269–76.

308. Hollingworth W, Todd C, Parker M, Roberts JA,Williams R. Cost-analysis of early discharge after hip fracture. BMJ 1993;307:903–6.

309. Morris S, McGuire A, McPherson K. The cost-effectiveness of raloxifene therapy forpostmenopausal women in the UK. London: City University, 1999.

310. O’Cathain A. Evaluation of a hospital at homescheme for the early discharge of patients withfractured neck of femur. J Public Health Med1994;16:205–10.

311. Parker MJ, Todd CJ, Palmer CR, Camilleri-Ferrante C, Freeman C, Laxton C, et al. Inter-hospital variations in length of hospital stayfollowing hip fracture. Age Ageing 1998;27:333–7.

312. Reid DM, Torgerson DJ. Pharmacoeconomic aspectsof intermittent cyclical etidronate therapy in thetreatment of postmenopausal and corticosteroid-induced osteoporosis. Rev Contemp Pharmacother1998;9:287–92.

313. Taylor R, Kirby B. Cost implications of cardiacrehabilitation in older patients. Coron Artery Dis1999;10:53–6.

314. Torgerson DJ, Dolan P. Prescribing by generalpractitioners after an osteoporotic fracture. Ann Rheum Dis 1998;57:378–9.

315. Torgerson DJ, Reid DM. The pharmacoeconomicsof hormone replacement therapy.PharmacoEconomics 1999;16:9–16.

316. Torgerson DJ, Reid DM. Osteoporosis preventionthrough screening: will it be cost-effective? BaillieresClin Rheumatol 1993;7:603–22.

317. Torgerson DJ, Donaldson C, Reid DM. Usingeconomics to prioritise research: a case study of randomized trials for the prevention of hipfractures due to osteoporosis. J Health Serv ResPolicy 1996;1:141–6.

318. Torgerson DJ, Donaldson C, Reid DM. Bonemineral density measurements: are theyworthwhile? J R Soc Med 1996;89:457–61.


125

319. Torgerson DJ, Gosden T, Reid DM. The economicsof osteoporosis prevention. Trends Endocrinol Metab1997;8:236–9.

320. Townsend J. Hormone replacement therapy:assessment of present use, costs, and trends. Br J Gen Pract 1998;48:955–8.

321. Townsend J, Buxton M. Cost-effectiveness scenarioanalysis for a proposed trial of hormonereplacement therapy. Health Policy 1997;39:181–94.

322. Withey C, Morris R, Beach R, Backhouse A.Outcome following fractured neck of femur –variation in acute hospital care or case mix? J Public Health Med 1995;17:429–37.

323. Netten A, Denett J, Knight J. Unit costs of healthand social care 1998. Canterbury: Personal SocialServices Research Unit, University of Kent; 1998.

324. GDP and GDP deflators at market prices. 2000.URL: www.hmtreasury.gov.uk/deflators/figures.html (July 2000).

325. Pathways Through Care Study Group. Pathwaysthrough care: the processes and outcomes ofhospital care for elderly people. Newcastle-upon-Tyne: University of Newcastle, Centre for HealthServices Research; 1996.

326. Netten A, Denett J. Unit costs of health and socialcare. Canterbury: Personal Social Services ResearchUnit, University of Kent; 1996.

327. Yuen P. OHE compendium of health statistics. 11thedition. London: Office of Health Economics; 1999.

328. de Laet CEDH, van Hout BA, Burger H, WeelAEAM, Hofman A, Pols HAP. Incremental cost ofmedical care after hip fracture and first vertebralfracture: the Rotterdam study. Osteoporos Int1999;10:66–72.

329. Chrischilles E, Shireman T, Wallace R. Costs andhealth effects of osteoporotic fractures. Bone1994;15:377–86.

330. Wolstenholme JL, Smith SJ, Whynes DK. The costsof treating breast cancer in the United Kingdom:implications for screening. Int J Technol Assess HealthCare 1998;14:277–89.

331. Dolan P, Torgerson DJ, Wolstenholme J. Costs ofbreast cancer treatment in the United Kingdom.Breast 1999;8:205–7.

332. Pickin DM, Payne JN, Haq IU, McCabe CJ, Ward SE,Jackson PR, et al. Statin therapy/ Co-A reductaseinhibitor treatment in the prevention of coronaryheart disease. Guidance note for purchasers. Reportno. 96/04. Sheffield: Trent Institute for HealthServices Research; 1996.

333. Piercy J, Pledger G. Estimating the resourceimplications of coronary heart disease in Newcastle.(Cost-effective purchasing occasional paper 2).York: University of York Health EconomicsConsortium; 1991.

334. England and Wales mid-1996 population estimates.2000. URL:www.statistics.gov.uk/statsbase/xsdataset.asp [July 2000].

335. Netten A, Curtis L. Unit costs of health and socialcare 2000. Canterbury: University of Kent; 2000.

336. Garton MJ, Torgerson DJ, Donaldson C, Russell IT,Reid DM. Recruitment methods for screeningprogrammes: trial of a new method within aregional osteoporosis study. BMJ 1992;305:82–4.

337. Sonnenberg FA, Beck JR. Markov models in medicaldecision making. Med Decis Making 1993;13:322–38.

338. Sox HC, Blatt MA, Higgins MC, Marton KI. Medical decision making. Boston: Butterworth-Heinemann; 1998.

339. O’Hagan A. Curve fitting and optimal design forprediction [with discussion]. J Roy Statist Soc B1978;40:1–42.

340. Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost-effectiveness in health and medicine. London:Oxford University Press; 1996.

341. National Institute for Clinical Excellence 2000.http://www.nice.org.uk/nice-web/

342. Raftery J. NICE: a faster access to moderntreatments? Analysis of guidance on healthtechnologies. BMJ 2001;232:1300–3.

343. Nevitt MC, Thompson DE, Black DM, Rubin SR,Ensrud K, Yates AJ, et al. Effect of alendronate onlimited-activity days and bed-disability days causedby back pain in post-menopausal women withexisting vertebral fractures. Arch Intern Med2000;160:77–85.

344. Johnell O, Kanis JA, de Laet C, Jonsson B,Zethraeus N, Oden A. Sequential changes in qualityof life after osteoporotic fractures. Osteoporos Int2002;13 Suppl 1: S70.

345. Kanis JA, Delmas P, Burckhardt P, Cooper C,Togerson D. Guidelines for diagnosis and manage-ment of osteoporosis. Osteoporos Int 1997;7:390–406.

Trials meeting the inclusioncriteria

Trials are listed alphabetically by their identifying‘name’; the reference number relates to theprincipal publication(s) for each trial, indicated by an asterisk (*). Note that most of the sub-sidiary references do not appear in the mainreference list.


References

126

Abellan Perez, 1995139

* Abellan Perez M, Bayina Garcia FJ, Calabozo M,Carpintero Benitez P, Figueroa Pedrosa M, Fernandez Crisostoma C, et al. [Multicenter comparativestudy of synthetic salmon calcitonin administered nasally in the treatment of established postmenopausalosteoporosis. In Spanish.] An Med Interna 1995;12:12–16.

Adami, 199593

Adami S, Baroni MC, Broggini M, Carratelli L, Caruso I,Gnessi L, et al. Treatment of postmenopausal osteo-porosis with continuous daily oral alendronate incomparison with either placebo or intranasal salmoncalcitonin. Osteoporos Int 1993;3 Suppl 3:S21–7.

* Adami S, Passeri M, Ortolani S, Broggini M, Carratelli L, Caruso I, et al. Effects of oral alendronateand intranasal salmon calcitonin on bone mass andbiochemical markers of bone turnover in post-menopausal women with osteoporosis. Bone1995;17:383–90.

Agrawal, 198161

*Agrawal R, Wallach S, Cohn S, Tessier M, Verch R,Hussain M, et al. Calcitonin treatment of osteoporosis.In: Calcitonin 1980: proceedings of an internationalsymposium; Milan, October 1980. Amsterdam: ExcerptaMedica; 1981 p.237–46.

Aloia, 1988117

* Aloia JF, Vaswani A, Yeh JK, Ellis K, Yasumura S, Cohn SH. Calcitriol in the treatment of postmenopausalosteoporosis. Am J Med 1988;84:401–8.

Arthur, 1990190

* Arthur RS, Piraino B, Candib D, Cooperstein L, Chen T, West C, et al. Effect of low-dose calcitriol andcalcium therapy on bone histomorphometry and urinary calcium excretion in osteopenic women. Miner Electrolyte Metab 1990;16:385–90.

Birkenhager, 1992164

Birkenhager JC, Birkenhager-Frenkel DH, Erdtsieck RJ,Kooy PPM, Pols HAP. Hormonal replacement therapywith and without nandrolone in postmenopausalosteoporosis. J Bone Miner Res 1991;6 Suppl 1:S278.

* Birkenhager JC, Erdtsieck RJ, Zeelenberg J, van Kuik C,van Veen LC, Birkenhager-Frenkel DH, et al. Cannandrolone add to the effect of hormonal replacementtherapy in postmenopausal osteoporosis? Bone Miner1992;18:251–65.

Bone, 199794

* Bone HG, Downs RW Jr, Tucci JR, Harris ST, Weinstein R, Licata AA, et al. Dose–responserelationships for alendronate treatment in osteoporoticelderly women. J Clin Endocrinol Metab 1997;82:265–74.

Weinstein RS, Bone H, Tucci J, et al. Alendronatetreatment of osteoporosis in elderly women. J Bone MinerRes 1994;9 Suppl 1:S144, 1994.

Brockstedt, 199673

* Brockstedt H, Melsen F, Mosekilde L. Double-blind,randomized study comparing the therapeutic effects of continuous and pulse dosing sodiummonofluorophosphate (NaMFPh) in spinal crushfracture osteoporosis [abstract]. Osteoporos Int 1996;6 Suppl 1:254.

Carfora, 199895

* Carfora E, Sergio F, Bellini P, Sergio C, Falco D,Zarcone, R. Effect of treatment of postmenopausalosteoporosis with continuous daily oral alendronate and the incidence of fractures. Gazzetta Med Ital Arch SciMed 1998;157(4):105–9.

Chesnut, 198396

* Chesnut CH 3rd, Ivey JL, Gruber HE, Matthews M, Nelp WB, Sisom K, et al. Stanozolol in postmenopausalosteoporosis: therapeutic efficacy and possible mech-anisms of action. Metab Clin Exp 1983;32:571–80.

Chesnut, 199596

* Chesnut CH 3rd, McClung MR, Ensrud KE, Bell NH,Genant HK, Harris ST, et al. Alendronate treatment ofthe postmenopausal osteoporotic woman: effect ofmultiple dosages on bone mass and bone remodeling.Am J Med 1995;99:144–52.

Clemmesen, 199797

* Clemmesen B, Ravn P, Zegels B, Taquet AN,Christiansen C, Reginster JY. A 2-year phase II study with 1-year of follow-up of risedronate (NE-58095) in postmenopausal osteoporosis. Osteoporos Int1997;7:488–95.

Cristallini, 1993128

* Cristallini S, Pedetti M, Donatelli C, Gregorio F,Filipponi P. [The effect of different “coherent”-typetherapeutic plans on bone mineral density and on theincidence of vertebral fractures in advanced osteo-porosis. In Italian.] Recent Prog Med 1993;84:336–45.

Dykman, 198463

* Dykman TR, Haralson KM, Gluck OS, Murphy WA,Teitelbaum SL, Hahn TJ, et al. Effect of oral 1,25-dihydroxyvitamin D and calcium on glucocorticoid-induced osteopenia in patients with rheumatic diseases.Arthr Rheum 1984;27:1336–43.


127

Ebeling, 199870

* Ebeling PR, Wark JD, Yeung S, Poon C, Salehi N,Nicholson GC, et al. Effects of calcitriol or calcium on bone mineral density and fractures in men withidiopathic osteoporosis. Osteoporos Int 1998;8 Suppl 3:114, 1998.

Ebrahim, 1997187

* Ebrahim S, Thompson PW, Baskaran V, Evans K.Randomized placebo-controlled trial of brisk walking in the prevention of postmenopausal osteoporosis. Age Ageing 1997;26:253–60.

Ellerington, 1996129

* Ellerington MC, Hillard TC, Whitcroft SI, Marsh MS,Lees B, Banks LM, et al. Intranasal salmon calcitonin for the prevention and treatment of postmenopausalosteoporosis. Calcif Tissue Int 1996;59:6–11.

Ettinger, 1999182

Cummings SR, Eckert S, Krueger KA, Grady D, Powles TJ, Cauley JA, et al. The effect of raloxifene on risk of breast cancer in postmenopausal women:results from the MORE randomized trial. JAMA1999;281:2189–97.*Ettinger B, Black DM, Mitlak BH, Knickerbocker RK,Nickelsen T, Genant H, et al. Reduction of vertebralfracture risk in postmenopausal women with osteo-porosis treated with raloxifene: results from a 3-yearrandomized clinical trial. JAMA 1999;282:637–45.

Falch, 1987191

Falch JA, Odegaard OR, Finnanger AM. 3 years’treatment with 1.25(OH)2 vitamin D3 does not reducebone loss or fracture rate in postmenopausal womenwith fracture of the distal forearm. In: Norman AW,Schaefer K, Grigoleit H, Herrath D, editors. Vitamin D.Chemical, biochemical and clinical update. Berlin:Walter de Gruyter; 1985. p.1004–5.* Falch JA, Odegaard OR, Finnanger AM, Matheson I.Postmenopausal osteoporosis: no effect of three yearstreatment with 1,25-dihydroxycholecalciferol. Acta MedScand 1987;221:199–204.

FIT98,99

* Black DM, Cummings SR, Karpf DB, Cauley JA,Thompson DE, Nevitt MC, et al. Randomised trial ofeffect of alendronate on risk of fracture in women withexisting vertebral fractures. Lancet 1996;348:1535–41.

Black DM, Reiss TF, Nevitt MC, Cauley J, Karpf D,Cummings SR. Design of the Fracture Intervention Trial.Osteoporos Int 1993;3 Suppl 3:S29–39.

Cummings SR, Black DM, Thompson DE. Alendronatereduces the risk of vertebral fractures in women withoutpre-existing vertebral fractures: results of the FractureIntervention Trial. J Bone Miner Res 1996;12 Suppl 1:S149.

* Cummings SR, Black DM, Thompson DE, Applegate WB, Barrett-Connor E, Musliner TA, et al.Effect of alendronate on risk of fracture in women withlow bone density but without vertebral fractures: resultsfrom the Fracture Intervention Trial. JAMA1998;280:2077–82.

Cummings SR, Black DM, Vogt TM. Changes in BMDsubstantially underestimate the anti-fracture effects ofalendronate and other antiresorptive agents. J BoneMiner Res 1996;11 Suppl 1:S102.

Ensrud KE, Black DM, Palermo L, Bauer DC, Barrett-Connor E, Quandt SA, et al. Treatment with alendronateprevents fractures in women at highest risk: results fromthe Fracture Intervention Trial. Arch Intern Med1997;157:2617–24.

Hochberg MC, Ross PD, Black D, Cummings SR, Genant HK, Nevitt MC, et al. Larger increases in bone mineral density during alendronate therapy areassociated with a lower risk of new vertebral fractures in women with postmenopausal osteoporosis. ArthrRheum 1999;42:1246–54.

Hochberg MC, Ross PD, Cummings SR, Black D,Musliner T, Nevitt MC, et al. Larger increases in bonemineral density with alendronate therapy are associatedwith lower risk of new vertebral fractures. Osteoporos Int1998;8 Suppl 3:13.

Nevitt MC, Thompson DE, Black DM, Rubin SM, Ensrud K, Yates J, et al. The effect of osteoporosistreatment on limitation of activity due to back pain. J Bone Miner Res 1997;12 Suppl 1:S144.

Nevitt MC, Thompson DE, Black DM, Rubin SR, Ensrud K, Yates AJ, et al. Effect of alendronate onlimited-activity days and bed-disability days caused byback pain in postmenopausal women with existingvertebral fractures. Arch Intern Med 2000;160:77–85.

Fujita, 199264

* Fujita T, Fukase M, Shimada T, Yamamoto H.Treatment of established osteoporosis with 1-alpha (OH) vitamin D3 and low dose intermittent elcatonin(eel calcitonin derivative). J Bone Miner Metab1992;10:37–40.

Fujita, 199358

* Fujita T, Orimo H, Inoue T, Kaneda K, Sakurai M,Morita R, et al. [Double-blind multicenter comparativestudy with alfacalcidol of etidronate disodium (EHDP)in involutional osteoporosis. In Japanese.] Rinsho Hyoka21:261–302.

Gallagher, 1989118

* Gallagher JC, Riggs BL, Recker RR, Goldgar D. Theeffect of calcitriol on patients with postmenopausalosteoporosis with special reference to fracture frequency.Proc Soc Exp Biol Med 1989;3:287–92.


References

128

Gallagher, 1990a119

*Gallagher JC, Goldgar D. Treatment of postmenopausalosteoporosis with high doses of synthetic calcitriol. A randomized controlled study. Ann Intern Med1990;113:649–55.

Gallagher, 1990b74

*Gallagher JC. Effect of fluoride in osteoporosis[abstract]. J Bone Miner Res 1990;5 Suppl 1:S230.

Gennari, 198565

*Gennari C, Chierichetti SM, Bigazzi S, Fusi L, Gonnelli S, Ferrara R, et al. Comparative effects on bonemineral content of calcium and calcium plus salmoncalcitonin given in two different regimes in post-menopausal osteoporosis. Curr Ther Res 1985;38:455–64.

Geusens, 1986163

*Geusens P, Dequeker J. Long-term effect of nandrolonedecanoate, 1 alpha-hydroxyvitamin D3 or intermittentcalcium infusion therapy on bone mineral content, bone remodeling and fracture rate in symptomaticosteoporosis: a double-blind controlled study. Bone Miner1986;1:347–57.

Geusens P, Dequeker J, Verstraeten A, Nijs J, vanHolsbeeck M. Bone mineral content, cortical thicknessand fracture rate in osteoporotic women after withdrawalof treatment with nandrolone decanoate, 1-alphahydroxyvitamin D3, or intermittent calcium infusions.Maturitas 1986;8:281–9.

Guañabens, 199675

*Guañabens N, Farrerons J, Perez-Edo L, Monegal A,Renau A, Roca M, et al. Comparison of the efficiency and safety of cyclic etidronate versus sodium fluoride in postmenopausal osteoporosis [abstract]. Osteoporos Int 1996;6 Suppl 1:255.

Gutteridge, 199380

*Gutteridge DH, Kent GN, Prince RL, Nicholson GC,Stewart GO, Jones CE, et al. Fluoride treatment ofosteoporosis: cyclical non-blinded or continuous blindedstudies? Osteoporos Int 1993;3 Suppl 1:215–17.

Gutteridge, 199679

Gutteridge DH, Kent GN, Prince RL, Nicholson GC,Stewart GO, Jones CE, et al. Fluoride treatment ofosteoporosis: cyclical non-blinded or continuous blindedstudies? Osteoporos Int 1993a;3 Suppl 1:215–17.

Gutteridge DH, Price RI, Prince RL, Jones CE, Stewart GO, Nicholson GC, et al. Fluoride ± estrogen inosteoporosis – spinal bone gain, lower limb bone lossand stress fractures. J Bone Miner Res 1993b;3 Suppl1:S341.

*Gutteridge DH, Drury PH, Stewart GO, Prince RL, Price RI, Retallack RW, et al. Cyclic Na fluoride + HRT inpostmenopausal osteoporosis – more vertebral fractureswith NaF: less with HRT and NaF [abstract]. OsteoporosInt 1996;6 Suppl 1:263.

Hansson, 1987167

*Hansson T, Roos B. The effect of fluoride and calciumon spinal bone mineral content: a controlled,prospective (3 years) study. Calcif Tissue Int1987;40:315–17.

Harris, 1999100

*Harris ST, Watts NB, Genant HK, McKeever CD,Hangartner T, Keller M, et al. Effects of risedronatetreatment on vertebral and nonvertebral fracturesin women with postmenopausal osteoporosis: arandomized controlled trial. JAMA 1999;282:1344–52.

Hizmetli, 1998130

*Hizmetli S, Elden H, Kaptanoglu E, Nacitarhan V,Kocagil S. The effect of different doses of calcitonin on bone mineral density and fracture risk in post-menopausal osteoporosis. Int J Clin Pract 1998;52:453–5.

Hodsman, 1997131

*Hodsman AB, Fraher LJ, Watson PH, Ostbye T, Stitt LW, Adachi JD, et al. A randomized controlled trialto compare the efficacy of cyclical parathyroid hormoneversus cyclical parathyroid hormone and sequentialcalcitonin to improve bone mass in postmenopausalwomen with osteoporosis. J Clin Endocrinol Metab1997;82:620–8.

Kleerekoper, 1991168

*Kleerekoper M, Peterson EL, Nelson DA, Phillips E,Schork MA, Tilley B, et al. A randomized trial of sodiumfluoride as a treatment for postmenopausal osteoporosis.Osteoporos Int 1991;1:155–61.

Leidig-Bruckner, 199776

*Leidig-Bruckner G, Eberwein S, Bruckner T, Wolf S,Ziegler R. Long term follow-up of lumbar and femoralbone mineral density and vertebral fractures in patientswith manifest postmenopausal osteoporosis after fluoridetherapy alone or combined with estrogens or anabolica.Exp Clin Endocrinol Diabetes 1997;105:A58.

Lems, 1997192

*Lems WF, Jacobs JW, Bijlsma JW, van Veen GJ, Houben HH, Haanen HC, et al. Is addition of sodiumfluoride to cyclical etidronate beneficial in the treatmentof corticosteroid induced osteoporosis? Ann Rheum Dis1997;56:357–63.


129

Liberman, 1995101

Devogelaer JP, Broll H, Correa-Rotter R, Cumming DC,de Deuxchaisnes CN, Geusens P, et al. Oral alendronateinduces progressive increases in bone mass of the spine,hip, and total body over 3 years in postmenopausalwomen with osteoporosis. Bone 1996;18:141–50.

Harris ST, Watts NB, Hosking D, Adami S, Quan H,Shapiro DE, et al. Treatment with alendronate preventsfractures in women with postmenopausal osteoporosis.Osteoporos Int 1996;6 Suppl 1:261.

Liberman UI, Hirsch LJ. Esophagitis and alendronate. N Engl J Med 1996;335:1069–70.*Liberman UA, Weiss SR, Broll J, Minne HW, Quan H,Bell NH, et al. Effect of oral alendronate on bonemineral density and the incidence of fractures inpostmenopausal osteoporosis. N Engl J Med1995;333:1437–43.

Meunier PJ. Oral alendronate increases bone-mineraldensity and reduces vertebral fracture incidence inpostmenopausal osteoporosis. Br J Rheumatol 1997;36 Suppl 1:15–19.

Recker RR, Karpf DB, Quan H, Devogelaer C, Leite MO,Farus MJ, et al. Three-year treatment of osteoporosis with alendronate: effect on vertebral fracture incidence.In: Proceedings 77th Annual Meeting, EndocrineSociety, 1995.

Seeman E, de Deuxchaisnes CN, Meunier P, Santora AC.Treatment of postmenopausal osteoporosis with oralalendronate. Bone 1995;16 Suppl 1:S120.

Tucci JR, Tonino RP, Emkey RD, Peverly CA, Kher U,Santora AC 2nd. Effect of three years of oral alendronatetreatment in postmenopausal women with osteoporosis.Am J Med 1996;101:488–501.

Lindsay, 1999102

*Lindsay R, Cosman F, Lobo RA, Walsh BW, Harris ST,Reagan JE, et al. Addition of alendronate to ongoinghormone replacement therapy in the treatment ofosteoporosis: a randomized, controlled clinical trial. J Clin Endocrinol Metab 1999;84:3076–81.

Lufkin, 1992143

*Lufkin EG, Wahner HW, O’Fallon WM, Hodgson SF,Kotowicz MA, Lane AW, et al. Treatment of post-menopausal osteoporosis with transdermal estrogen. Ann Intern Med 1992;117:1–9.

Lufkin, 1998178

Nickelsen T, Lufkin EG, Riggs BL, Cox DA, Crook TH.Raloxifene hydrochloride, a selective estrogen receptormodulator: safety assessment of effects on cognitivefunction and mood in postmenopausal women.Psychoneuroendocrinology 1999;24:115–28.*Lufkin EG, Whitaker MD, Nickelsen T, Argueta R,Caplan RH, Knickerbocker RK, et al. Treatment ofestablished postmenopausal osteoporosis with raloxifene:a randomized trial. J Bone Miner Res 1998;13:1747–54.

Lyritis, 1994166

*Lyritis GP, Androulakis C, Magiasis B, Charalambaki Z,Tsakalakos N. Effect of nandrolone decanoate and 1-alpha-hydroxy-calciferol on patients with vertebralosteoporotic collapse. A double-blind clinical trial. Bone Miner 1994;27:209–17.

Mamelle, 1988176

*Mamelle N, Meunier PJ, Dusan R, Guillaume M, Martin JL, Gaucher A, et al. Risk–benefit ratio of sodiumfluoride treatment in primary vertebral osteoporosis.Lancet 1988;ii:361–5.

Mamelle N, Meunier PJ, Netter P. Fluoride and vertebralfractures. Lancet 1990;336:243.

Maugeri, 1994157

*Maugeri D, Panebianco P, Russo MS, Motta M, Tropea S, Motta L, et al. Ipriflavone-treatment of senileosteoporosis: results of a multicenter, double-blindclinical trial of 2 years. Arch Gerontol Geriatr1994;19:253–63.

McClung, 199877

McClung M, Bensen W, Bolognese M, Bonnick S,Ettinger M, Harris S, et al. Risedronate treatment ofpostmenopausal women with low bone mass: preliminarydata [abstract]. Osteoporos Int 1996;6 Suppl 1:257.*McClung M, Bensen W, Bolognese M, Bonnick S,Ettinger M, Harris S, et al. Risedronate increases bonemineral density at the hip, spine and radius inpostmenopausal women with low bone mass. Osteoporos Int 1998;8 Suppl 3:111.

Meunier, 1998169

*Meunier PJ, Sebert JL, Reginster JY, Briancon D,Appelboom T, Netter P, et al. Fluoride salts are no better at preventing new vertebral fractures thancalcium-vitamin D in postmenopausal osteoporosis: the FAVO study. Osteoporos Int 1998;8:4–12.

Montessori, 1997103

*Montessori ML, Scheele WH, Netelenbos JC, Kerkhoff JF, Bakker, K. The use of etidronate andcalcium versus calcium alone in the treatment ofpostmenopausal osteopenia: results of three years oftreatment. Osteoporos Int 1997;7:52–8.

Orimo, 1987120

*Orimo H, Shiraki M, Hayashi T, Nakamura, T. Reduced occurrence of vertebral crush fractures insenile osteoporosis treated with 1 alpha (OH)-vitaminD3. Bone Miner 1987;3:47–52.


References

130

Orimo, 1994121

*Orimo H, Shiraki M, Hayashi Y, Hoshino T, Onaya T,Miyazaki S, et al. Effects of 1 alpha-hydroxyvitamin D3 on lumbar bone mineral density and vertebral fracturesin patients with postmenopausal osteoporosis. CalcifTissue Int 1994;54:370–6.

Ott, 1989122

*Ott SM, Chesnut CH 3rd. Calcitriol treatment is not effective in postmenopausal osteoporosis [seecomments]. Ann Intern Med 1989;110:267–74.

Ott SM, Chesnut CH 3rd. Tolerance to doses of calcitriol is associated with improved bone density inwomen with postmenopausal osteoporosis. J Bone MinerRes 1990;5:S186.

Overgaard, 199260

Overgaard K, Christiansen C. A new biochemical markerof bone resorption for follow-up on treatment with nasalsalmon calcitonin. Calcif Tissue Int 1996;59:12–16.*Overgaard K, Hansen MA, Jensen SB, Christiansen C.Effect of salcatonin given intranasally on bone mass andfracture rates in established osteoporosis: a dose-response study. BMJ 1992;305:556–61.

Pacifici, 1988104

*Pacifici R, McMurtry C, Vered I, Rupich R, Avioli LV.Coherence therapy does not prevent axial bone loss inosteoporotic women: a preliminary comparative study. J Clin Endocrinol Metab 1988;66:747–53.

Pak, 1989193

*Pak CY, Sakhaee K, Zerwekh JE, Parcel C, Peterson R,Johnson K. Safe and effective treatment of osteoporosiswith intermittent slow release sodium fluoride:augmentation of vertebral bone mass and inhibition of fractures. J Clin Endocrinol Metab 1989;68:150–9.

Pak, 1995170

*Pak CY, Sakhaee K, Adams-Huet B, Piziak V, Peterson RD, Poindexter JR. Treatment of post-menopausal osteoporosis with slow-release sodiumfluoride. Final report of a randomized controlled trial.Ann Intern Med 1995;123:401–8.

Pak CY, Sakhaee K, Piziak V, Peterson RD, Breslau NA,Boyd P, et al. Slow-release sodium fluoride in the man-agement of postmenopausal osteoporosis. A randomizedcontrolled trial [see comments]. Ann Intern Med1994;120:625–32.

Passeri, 1992158

*Passeri M, Biondi M, Costi D, Bufalino L, Castiglione GN, Di Peppe C, et al. Effect of ipriflavoneon bone mass in elderly osteoporotic women. Bone Miner1992;19 Suppl 1:S57–62.

Passeri, 1995159

*Passeri M, Biondi M, Costi D, Dall’Aglio E, Pedrazzoni M, Bufalino L, et al. Effects of 2-year therapywith ipriflavone in elderly women with establishedosteoporosis. Ital J Miner Electrolyte Metab 1995;9:137–44.

Pols, 1999105

*Pols HA, Felsenberg D, Hanley DA, Stepan J, Munoz-Torres M, Wilkin TJ, et al. Multinational, placebo-controlled, randomized trial of the effects of alendronate on bone density and fracture risk inpostmenopausal women with low bone mass: results of the FOSIT study. Osteoporos Int 1999;9:461–8.

Pontiroli, 1991132

*Pontiroli AE, Pajetta E, Calderara A, Alberetto M, Pozza G, Manganelli V, et al. Intranasal and intra-muscular human calcitonin in female osteoporosis andin Paget’s disease of bones: a pilot study. J EndocrinolInvest 1991;14:47–51.

‘PROOF’ study78

Chesnut CH III. Salmon calcitonin in the treatment ofosteoporosis. Osteoporos Int 1998;8 Suppl 3:147.*Chesnut C, Baylink DJ, Doyle D, Genant H, Harris S,Kiel DP, et al. Salmon calcitonin nasal spray preventsvertebral fractures in established osteoporosis: furtherinterim results of the ‘PROOF’ study. Osteoporos Int1998;8 Suppl 3:13.

Gimona A. The PROOF study: methodology. OsteoporosInt 1998;8 Suppl 3:146.

Kanis JA, McCloskey EV. Effect of calcitonin on vertebraland other fractures. QJM 1999;92:143–9.

Stock JL, Avioli LV, Baylink DJ, Chesnut C, Genant HK,Maricic MJ, et al. Calcitonin–salmon nasal spray reducesthe incidence of new vertebral fractures in post-menopausal women: three year interim results of the PROOF study. J Bone Miner Res 1997;12:S149.

Recker, 1996142

*Recker RR, Hinders S, Davies KM, Heaney RP, Stegman MR, Lappe JM, et al. Correcting calciumnutritional deficiency prevents spine fractures in elderly women. J Bone Miner Res 1996;11:1961–6.

Recker R, Kimmel DB, Hinders S, Davies KM.Antifracture efficacy of calcium in elderly women. J Bone Miner Res 1994;9 Suppl 1:135.

Reginster, 1998171

*Reginster JY, Meurmans L, Zegels B, Rovati LC, Minne HW, Giacovelli G, et al. The effect of sodiummonofluorophosphate plus calcium on vertebral fracture rate in postmenopausal women with moderateosteoporosis. A randomized, controlled trial. Ann Intern Med 1998;129:1–8.


131

Reginster JY, Zegels B, Meurmans L, Rovati LC, Taquet AN, Setnikar I, et al. Monofluorophosphatedecreases vertebral fracture rate in postmenopausalosteoporosis: a randomised, placebo-controlled, doubleblind study. Osteoporos Int 1996;6 Suppl 1:239.

Reginster, 2000106

*Reginster J, Minne HW, Sorensen OH, Hooper M, Roux C,Brandi ML, et al. Randomized trial of the effects ofrisedronate on vertebral fractures in women with establishedpostmenopausal osteoporosis. Osteoporos Int 2000;11:83–91.

Reid, 1994107

*Reid IR, Wattie DJ, Evans MC, Gamble GD, Stapleton JP,Cornish J. Continuous therapy with pamidronate, apotent bisphosphonate, in postmenopausal osteoporosis.J Clin Endocrinol Metab 1994;79:1595–9.

Rico, 1992134

*Rico H, Hernandez ER, Revilla M, Gomez-Castresana F.Salmon calcitonin reduces vertebral fracture rate in post-menopausal crush fracture syndrome. Bone Miner1992;16:131–8.

Rico, 1995135

*Rico H, Revilla M, Hernandez ER, Villa LF, Alvarez deBuergo M. Total and regional bone mineral content andfracture rate in postmenopausal osteoporosis treatedwith salmon calcitonin: a prospective study. Calcif TissueInt 1995;56:181–5.

Riggs, 1990172

Melton LJ 3rd, Egan KS, O’Fallon WM, Riggs BL.Influence of fracture criteria on the outcome of arandomized trial of therapy. Osteoporos Int 1998;8:184–91.*Riggs BL, Hodgson SF, O’Fallon WM, Chao EY, Wahner HW, Muhs JM, et al. Effect of fluoride treatmenton the fracture rate in postmenopausal women withosteoporosis. N Engl J Med 1990;322:802–9.

Ringe, 1987136

*Ringe JD, Welzel D. Salmon calcitonin in the therapy of corticoid-induced osteoporosis. Eur J Clin Pharmacol1987;33:35–9.

Ringe, 1990137

*Ringe JD. [Treatment of primary osteoporosis withcalcium and salmon calcitonin. In German]. Dtsch MedWochenschr 1990;115:1176–82.

Ringe, 1998173

*Ringe JD, Dorst A, Kipshoven C, Rovati LC, Setnikar I.Avoidance of vertebral fractures in men with idiopathicosteoporosis by a three year therapy with calcium andlow-dose intermittent monofluorophosphate. OsteoporosInt 1998;8:47–52.

Ringe JD, Kipshoven C, Rovati L, Setnikar I. Therapy of idiopathic male osteoporosis: a three year study with calcium and low dose intermittent monofluoro-phosphate [abstract]. Osteoporos Int 1996;6 Suppl 1:96.

Ringe, 1999174

*Ringe JD, Kipshoven C, Coster A, Umbach R. Therapy of established postmenopausal osteoporosiswith monofluorophosphate plus calcium: dose-relatedeffects on bone density and fracture rate. Osteoporos Int1999;9:171–8.

Rozhinskaya, 199972

*Rozhinskaya L, Marova E, Sazonova N. Effectiveness of monofluorophosphate in established steroidosteoporosis. Osteoporos Int 1999;9(4 suppl 1):S11–12.

Schurch, 1998186

Schurch MA, Rizzoli R, Slosman D, Bonjour JP. Protein supplements increase serum IGF-1 and preventproximal femur bone loss in elderly with a recent hipfracture. Osteoporos Int 1996;6 Suppl 1:94.

*Schurch MA, Rizzoli R, Slosman D, Vadas L, Vergnaud P, Bonjour JP. Protein supplements increaseserum insulin-like growth factor-I levels and attenuateproximal femur bone loss in patients with recent hipfracture. A randomized, double-blind, placebo-controlled trial. Ann Intern Med 1998;128:801–9.

Sebert, 1995175

Sebert JL, Richard P, Mennecier I, Bisset JP, Loeb G. Adouble-blind study with a combined therapy (mono-fluorophosphate and calcium) in marked osteopeniawithout vertebral fracture. J Bone Miner Res 1993;8 Suppl 1:S255.

*Sebert JL, Richard P, Mennecier I, Bisset JP, Loeb G.Monofluorophosphate increases lumbar bone density inosteopenic patients: a double-masked randomized study.Osteoporos Int 1995;5:108–14.

Shiota, 1998194

*Shiota E. Evaluation of the drug therapy for establishedosteoporosis by dual-energy x-ray absorptiometry.Fukuoka Igaku Zasshi 1998;89:172–8.

Shiraki, 1996123

*Shiraki M, Kushida K, Yamazaki K, Nagai T, Inoue T,Orimo H. Effects of 2 years’ treatment of osteoporosiswith 1 alpha-hydroxy vitamin D3 on bone mineraldensity and incidence of fracture: a placebo-controlled,double-blind prospective study. Endocr J 1996;43:211–20.


References

132

Shiraki, 1999195

*Shiraki M, Kushid K, Fukunaga M, Kishimoto H, Taga M, Nakamura T, et al. A double-masked multicentercomparative study between alendronate and alfacalcidolin Japanese patients with osteoporosis. Osteoporos Int1999;10:183–92.

Shiraki, 200068

*Shiraki M, Shiraki Y, Aoki C, Miura M. Vitamin K2(menatetrenone) effectively prevents fractures andsustains lumbar bone mineral density in osteoporosis. J Bone Miner Res 2000;15:515–21.

Storm, 1990108

*Storm T, Thamsborg G, Steiniche T, Genant HK,Sorensen OH. Effect of intermittent cyclical etidronatetherapy on bone mass and fracture rate in women withpostmenopausal osteoporosis. N Engl J Med1990a;322:1265–71.

Storm T, Thamsborg G, Steiniche T, Genant HK,Sorensen OH. Etidronate for postmenopausalosteoporosis [letter]. N Engl J Med 1990b;323:1210.

Thiébaud, 1994115

*Thiébaud D, Burckhardt P, Melchior J, Eckert P, Jacquet A, Schnyder P, et al. Two years’ effectiveness ofintravenous pamidronate (APD) versus oral fluoride for osteoporosis occurring in the postmenopause.Osteoporos Int 1994;4:76–83.

Tilyard, 1992124

Tilyard M. Low-dose calcitriol versus calcium inestablished postmenopausal osteoporosis. Metabolism1990a;39(4 suppl 1):50–2.

Tilyard MW. 1,25 dihydroxyvitamin-d3 vs calcium in thetreatment of established postmenopausal osteoporosis. J Bone Miner Res 1990b;5 Suppl 2:S275.

Tilyard MW. 1,25 dihydroxyvitamin-d3 (calcitriol) in the treatment of postmenopausal osteoporosis. In:Cohn DV, Gennari C, Tashjian AH Jr, editors. Calciumregulating hormones and bone metabolism: basic and clinical aspects. volume II. Amsterdam: ElsevierScience; 1992. p.395–9.*Tilyard MW, Spears GF, Thomson J, Dovey S. Treatmentof postmenopausal osteoporosis with calcitriol orcalcium. N Engl J Med 1992;326:357–62.

Tilyard MW. Treatment of postmenopausal osteoporosiswith 1,25-dihydroxyvitamin D3. Osteoporos Int 1993;3Suppl 1:194–5.

Watts, 1990109

Harris ST, Watts NB, Jackson RD, Genant HK, WasnichRD, Ross P, et al. Four-year study of intermittent cyclicetidronate treatment of postmenopausal osteoporosis:three years of blinded therapy followed by one year ofopen therapy. Am J Med 1993;95:557–67.

Harris ST. Four-year study of intermittent cyclicetidronate treatment of postmenopausal osteoporosis[letter]. Am J Med 1995;99:225.

Miller PD, Watts NB, Licata AA, Harris ST, Genant HK,Wasnich RD, et al. Cyclical etidronate in the treatment ofpostmenopausal osteoporosis: efficacy and safety afterseven years of treatment. Am J Med 1997;103:468–76.*Watts NB, Harris ST, Genant HK, Wasnich RD, MillerPD, Jackson RD, et al. Intermittent cyclical etidronatetreatment of postmenopausal osteoporosis. N Engl J Med1990;323:73–9.

Wimalawansa, 1998110

*Wimalawansa SJ. A four-year randomized controlledtrial of hormone replacement and bisphosphonate,alone or in combination, in women with post-menopausal osteoporosis. Am J Med 1998;104:219–26.

Zarcone, 1997144

*Zarcone R, Carfora E, Sergio F, Bellini P, Longo M,Tartaglia E, et al. [Oestrogen therapy in women withpostmenopausal osteoporosis. In Italian]. MinervaGinecol 1997;49:355–9.

Trials excluded from thesystematic reviewThese are listed in alphabetical order, with anyreference number appearing at the end of thereference in parentheses [ ]. The reason forexclusion is given after each reference.

Buckle RM. 3 year study of sodium fluoride treatment onvertebral fracture incidence in osteoporosis. J Bone MinerRes 1989;4:S186. [62]

(Described as double-blind but does not specify thatpatients were randomised to treatment groups.)

Caniggia A, Delling G, Nuti R, Lore F, Vattimo A.Clinical, biochemical and histological results of a double-blind trial with 1,25-dihydroxyvitamin D3, estradiol andplacebo in post-menopausal osteoporosis. Acta VitaminolEnzymol 1984;6:117–28.


Chevalley T, Rizzoli R, Nydegger V, Slosman D, Rapin CH, Michel JP, et al. Effects of calciumsupplements on femoral bone mineral density andvertebral fracture rate in vitamin-D-replete elderlypatients. Osteoporos Int 1994;4:245–52.

(Compared two different calcium preparations butcomparative data on their effects on fracture rates not provided.)


133

Devogelaer JP, Boutsen Y, Malghem J, Depresseux G,Nagant de Deuxchaisnes C. Efficacy of therapy ofinvolutional osteoporosis with disodium mono-fluorophosphate and calcium. Osteoporos Int 1996;6 Suppl 1:252.


Diamond T, McGuigan L, Schone M, Levy S, Rae D. A 2 year open randomized controlled trial comparingcalcitriol to cyclical etidronate for the treatment ofglucocorticoid- induced osteoporosis. J Bone Miner Res1997;12:S634.

(Unavailable within the study timescale.)

Dilsen G, Gülbaba G, Sindel D. Calcitriol in thetreatment of osteoporosis. Osteoporos Int 1998;8 Suppl 3:101. [69]

(Results in the intervention and control arms not compared.)

Gallagher JC. Treatment of postmenopausal osteoporosiswith fluoride plus either calcium or calcitriol. J BoneMiner Res 1992;7 Suppl 1:S318. [71]

(Results in the intervention and control arms notcompared.)

Hayashi Y. Investigation of fracture frequency ofJapanese osteoporosis patients: effects of drug therapy. J Bone Miner Metab 1988;6:120. [66]

(Unavailable within the study timescale.)

Hayashi Y, Fujita T, Inoue T. Decrease of vertebralfracture in osteoporotics by administration of 1 alpha hydroxy-vitamin D3. J Bone Miner Metab1992;10:184–8. [67]

(Both men and women were quasi-randomised, by alternate allocation, but only results for women presented.)

Hedlund LR, Gallagher JC. Increased incidence of hipfracture in osteoporotic women treated with sodiumfluoride. J Bone Miner Res 1989;4:223–5.

(Patients not randomised to two groups as reported:results of two separate randomised trials are combined –one comparing sodium fluoride plus calcium withsodium fluoride plus calcitriol, the other comparingcalcitriol with placebo – in order to compare effects of fluoride with those of calcitriol/placebo.)

Itami Y, Fujita T, Inoue T, et al. Effect of alphacalcidol onosteoporosis – multicenter double-blind study. J Clin ExpMed 1982;123:958–73.

(Unavailable within study timescale.)

Lyritis GP, Tsakalakos N, Paspati I, Skarantavos G,Galanos A, Androulakis C. The effect of a modifiedetidronate cyclical regimen on postmenopausalosteoporosis: a four-year study. Clin Rheum1997;16:354–60.

(Although specified that patients were enrolled atrandom, it was not clearly specified that they wererandomised to treatment groups.)

Meunier PJ. Calcium, vitamin D and vitamin K in theprevention of fractures due to osteoporosis. Osteoporos Int1999;9 (Suppl 2):48–52.

(Unavailable within study timescale.)

Nakatsuka K, Inaba M, Aratani H, Iba K, Sato T, Koike T,et al. [Effects of long-term administration of alfacalcidolon bone mass and bone metabolism in patients withprimary osteoporosis – comparison with calciumpreparations. In Japanese]. Nippon Ronen Igakkai Zasshi 1997;34:569–76. [57]

(Unable to obtain translation.)

Ringe JD, Coster A, Meng T, Schacht E, Umbach R.Treatment of glucocorticoid-induced osteoporosis withalfacalcidol/calcium versus vitamin D/calcium. CalcifTissue Int 1999;65:337–40.

(Not specified that patients randomised into studygroups.)

Shiraki M [Vitamin K2. In Japanese]. Nippon Rinsho1998;56:1525–30. [56]

(Unable to obtain translation.)

Vose GP, Keele DK, Milner AM, Rawley R, Roach TL,Sprinkle EE 3rd. Effect of sodium fluoride, inorganicphosphate, and oxymetholone therapies in osteoporosis:a six-year progress report. J Gerontol 1978;33:204–12.

(Some patients had normal BMD.)



135

1 exp osteoporosis/2 bone diseases, metabolic/3 osteoporo$.tw.4 1 or 2 or 35 (bone adj6 densit$).tw.6 bone density/7 (bone or bones).mp.8 exp densitometry/9 tomography, x-ray computed/10 densit$.tw.11 9 and 1012 8 or 1113 7 and 1214 5 or 6 or 1315 Colles’ fracture/16 exp hip fractures/17 spinal fractures/18 15 or 16 or 1719 fractures/20 colles$.tw.21 (hip or hips).tw.22 (femur adj6 neck).tw.23 (femoral adj6 neck).tw.24 (spine or spinal).tw.25 vertebra$.tw.26 lumbar vertebrae/27 20 or 21 or 22 or 23 or 24 or 25 or 2628 19 and 2729 fractur$.tw.30 20 or 21 or 22 or 23 or 24 or 2531 (fractur$ adj6 (colles$ or (hip or hips) or

(femur adj6 neck) or (femoral adj6 neck) or(spine or spinal) or vertebra$)).tw.

32 18 or 28 or 3133 estrogen replacement therapy/34 estrogen replacement therapy.tw.35 oestrogen replacement therapy.tw.36 hormone replacement therapy.tw.37 ert.tw.38 ort.tw.39 hrt.tw.40 33 or 34 or 35 or 36 or 37 or 38 or 39

41 exp menopause/42 climacteric/43 menopaus$.tw.44 postmenopaus$.tw.45 climacteric.tw.46 41 or 42 or 43 or 44 or 4547 40 or 4648 32 and 4749 4 or 14 or 4850 randomized controlled trial.pt.51 controlled clinical trial.pt.52 randomized controlled trials/53 random allocation/54 double-blind method/55 single-blind method/56 50 or 51 or 52 or 53 or 54 or 5557 (animal not human).sh.58 56 not 5759 clinical trial.pt.60 exp clinical trials/61 (clin$ adj25 trial$).tw.62 ((singl$ or doubl$ or trebl$ or tripl$) adj25

(blind$ or mask$)).tw.63 placebos/64 placebo$.tw.65 random$.tw.66 research design/67 59 or 60 or 61 or 62 or 63 or 64 or 65 or 6668 67 not 5769 68 not 5870 comparative study.sh.71 evaluation studies/72 follow-up studies/73 prospective studies/74 (control$ or prospectiv$ or volunteer$).tw.75 70 or 71 or 72 or 73 or 7476 75 not 5777 76 not (58 or 69)78 58 or 69 or 7779 49 and 78


Appendix 1

MEDLINE search strategy for RCTs


137

The five journals identified for handsearchingwere as follows:

• Osteoporosis International• Journal of Bone and Mineral Research• Annals of Internal Medicine• New England Journal of Medicine• American Journal of Medicine.

It was not possible to handsearch the entire runs of these journals from January 1990 onwardsas had been intended, because volumes were either missing from local libraries or could not be supplied by the British Library. Thus, only the issues listed below were searched.

• Osteoporosis International: 1990, 1(1); 1994,5(1–6); 1995, 6(1–6 plus suppl 1 & 2); 1996,7(1–6 plus suppl 1, 2 & 3); 1997, 8(1–6 plussuppl 1 & 3); 1998, 9(1–2, 4–5); 1999, 10(1).

• Journal of Bone and Mineral Research: 1990, 5;1991, 6 (incl suppl 1); 1992, 7(3–4,7,9–10,12plus suppl 1); 1993, 8(1–4,7–12, plus suppl 1 & 2); 1998, 13; 1999, 14(1–5,8).

• Annals of Internal Medicine: 1990–99, 112–130;2000, 132(1–9).

• New England Journal of Medicine: 1990–97,322–337; 1998–2000, 339–341, 342(1–19).

• American Journal of Medicine: 1990–97, 88–103;1998, 104(1–4,6 plus suppl 2A,3A,4A,5A), 105;1999, 106(1–4,6, plus suppl 1A,5A,5B), 107.


Appendix 2

Details of handsearching


139


Appendix 3

Quality assessment tool*

Score

Was randomisation to study groups blinded?Not randomised 0

States randomised but no description or quasi-randomised (i.e. allocation by date of birth, hospital 1record number, admission dates, alternately, etc)

Small but real chance of disclosure of assignment (e.g. sealed envelopes) 2

Method does not allow disclosure of assignment (e.g. assigned by telephone communication, or by 3indistinguishable drug treatments randomly precoded by centralised pharmacy)

Were assessors of outcome blinded to treatment status?Not mentioned 1

Moderate chance of unblinding of assessors 2

Action taken to blind assessors, or outcomes such that bias is unlikely 3

Were outcomes of patients who withdrew described and included in analysis?Not mentioned or states number of withdrawals only 1

Stated numbers and reasons for withdrawal but analysis unmodified 2

Primary analysis based on all cases as randomised 3

Comparability of treatment and control groups at entry

Large potential for confounding or not discussed 1

Confounding small: mentioned but not adjusted for 2

Unconfounded: good comparability of groups or confounding adjusted for 3

For hip or other appendicular skeleton fractureNot applicable 0

No confirmation of diagnosis 1

X-ray confirmation of diagnosis 2

For vertebral fractureNot applicable 0

Inadequately described method 1

Radiological method: uses anterior/posterior height ratio 2

Radiological method: uses anterior, middle and posterior height in criteria OR reports radiologically 3confirmed clinical events only

Total methodology score (actual score as percentage of possible score)

* After: Gillespie WJ, Henry DA, O’Connell DL, Robertson J. Vitamin D and vitamin D analogues for preventingfractures associated with involutional and post-menopausal osteoporosis [Cochrane review]. The Cochrane Library,Issue 1; Oxford: Update Software; 1999


141

Bisphosphonates.................................................. 142

Vitamin D derivatives .......................................... 158

Calcitonin ............................................................ 164

Calcium ................................................................ 176

Oestrogen ............................................................ 178

Oestrogen-like molecules .................................. 182

Anabolic steroids ................................................ 185

Fluoride................................................................ 186

SERMs .................................................................. 196

Exercise ................................................................ 198

Protein supplements .......................................... 199

Vitamin K2 ............................................................ 200

Comparisons with active treatments .................. 201


Appendix 4

Details of RCTs (by study)*

* NB. The reference numbers quoted are the principal ones for each particular trial. For details of other relevantreferences, see ‘Trials meeting the inclusion criteria’ (page 125). This applies also to the references cited as sources of ‘Additional information’.

Appendix 4

142

Bisphosphonates

Study Adami, 199593

Setting Italy

Date of intervention Not specified

Source of funding Not specified

Design RCT, double-blind between alendronate and placebo, with open-label calcitonin arm

Study population Postmenopausal women with osteoporosis (lumbar T-score > –2), 5% of whom had vertebral fracture at entry

Recruitment procedure used Not specified

Number of patients 286

Length of study 2 years

Main intervention/s AlendronateIntranasal salmon calcitonin

Primary outcome measures BMD (spine)

Secondary outcome measures BMD (hip)Biochemical indices of bone turnover

Definition of incident Not applicable: only clinically apparent fractures were recordedvertebral fracture

Results: all fractures No significant trends noted between treatment groups in relation to number of fractures

Quality score 9/15

Comments • Alendronate arm placebo-controlled but calcitonin arm not because no intranasal placebo available• Blinding not possible in relation to calcitonin arm and although assessors of BMD scans were blinded to

treatment allocation, no such assurance is given in relation to fracture outcomes• All patients received elemental calcium, 500 mg daily• 41 women (14.3%) withdrew from treatment, 15 (5.2%) because of adverse events• Analysis was on ITT basis• All treatment groups were similar to placebo with regard to both the overall safety profile and

upper gastrointestinal adverse events• Quality score low partly because of lack of information about method of randomisation, but more

because of lack of information about methods used to identify and confirm fractures. However, study not designed to detect significant differences in fracture rates, and fracture data only collected as part of adverse event reporting. Perhaps, therefore, unreasonable to expect such detail to be provided

Study Bone, 199794

Setting USA


Source of funding Merck Research Laboratories

Design Multicentre, randomised, double-blind, placebo-controlled

Study population Healthy elderly women with osteopaenia or osteoporosis (T-score < –2 but no more than one lumbar crush fracture): 37% had vertebral fracture at entry

Recruitment procedure used ‘At 15 clinical sites throughout the United States’



Main intervention/s Alendronate

Primary outcome measures Lumbar BMD

Secondary outcome measures Biochemical indices of bone and mineral metabolismBone histomorphometryVertebral and non-vertebral fractures

Definition of incident 20% or more decrease in vertebral height*

vertebral fracture

Results: vertebral fracture Alendronate did not have a statistically significant effect on vertebral fracture

Results: non-vertebral fracture At doses of 2.5 and 5.0 mg, alendronate significantly reduced incidence of non-vertebral fracture

Quality score 14/18

Comments • All patients received elemental calcium, 500 mg daily• 131 women (36.5%) withdrew, 38 (10.6%) because of adverse events• No significant difference between treatment and placebo groups in terms of adverse effects which were

suspected to be drug-related: 19.8% of women on alendronate, 1.0 mg, 25.8% on 2.5 mg, 17.2% on 5.0 mg, and 23.1% in placebo group suffered such adverse effects

• Alendronate as well-tolerated by women aged over 70 years as by those aged 60–69 years

* Method described in Genant HK, et al. J Bone Miner Res 1993;8:1137–48.


143


Bisphosphonates contd

Study Carfora, 199895

Setting Italy

Date of intervention December 1993–May 1996


Design Randomised, placebo-controlled

Study population Postmenopausal women with osteoporosis (lumbar spine T-score < –2.5)



Length of study 30 months


Outcome measures BMDVertebral fracturesBiochemical markers

Definition of incident Definition not givenvertebral fracture

Results: vertebral fracture RR = 0.55 for new fracture in women treated with alendronate, compared with those receiving placebo

Results: non-vertebral fracture –

Quality score 7/15

Comments • All patients received elemental calcium, 500 mg daily• No information provided regarding withdrawals• Tolerance said to be excellent, the only adverse events being intolerance ‘at superior tract of the

gastroenteric apparatus’ and cutaneous rash. Episodes of nausea, dyspepsia, mild gastro-oesophagitis and abdominal pain appeared during first 15 months of treatment with alendronate, 20 mg

Study Chesnut, 199596

Setting USA




Study population Healthy postmenopausal white or Asian women with spinal osteopenia (lumbar spine BMD 0.88 g/cm2 or less) but no vertebral or hip fractures attributable to osteoporosis

Recruitment procedure used Advertisements and medical announcements





Secondary outcome measures Vertebral fractureNon-vertebral fracture


Results: vertebral fracture No vertebral fractures in any patient

Results: non-vertebral fracture 13 non-vertebral fractures occurred in 12 patients, evenly distributed across treatment groups and not considered related to therapy

Quality score 9/18

Comments • All patients received elemental calcium, 500 mg daily• 34 women (18.1%) withdrew during course of study. No information given as to how many withdrew

from treatment and control arms• 18 women withdrew because of adverse clinical experiences, of whom nine were women in

alendronate arm who withdrew because of adverse upper gastrointestinal events (seven receiving 40 mg and only one < 20 mg daily); another withdrew because of a rash which was considered to be alendronate-related

• Generally, alendronate was associated with few side-effects• Study not designed to identify effect of alendronate on skeletal fractures• Presence of vertebral fractures attributable to osteoporosis was exclusion criterion and may

explain total absence of incident vertebral fractures

Appendix 4

144


Study Clemmesen, 199797

Setting Belgium and Denmark

Date of intervention December 1990–


Design Two-centre, double-masked, placebo-controlled randomised

Study population Healthy postmenopausal women with established osteoporosis (at least one but no more than four vertebral fractures)

Recruitment procedure used Mainly from outpatients attending two osteoporosis clinics


Length of study 2 years + 1 additional year of follow-up

Main intervention/s Oral risedronate, taken either continuously or cyclically for 2 weeks of 12-week cycle

Primary outcome measures BMD at spine

Secondary outcome measures BMD at femoral neck, trochanter and Ward’s triangleBiochemical markers of bone turnoverIncident vertebral fractures

Definition of incident A reduction of at least 15% (Belgium) or 25% (Denmark) in anterior-to-posterior wall ratio, orvertebral fracture in anterior or posterior wall compared with adjacent vertebrae

Results: vertebral fracture Tendency towards lower incidence and rate of new vertebral fractures in group taking daily continuous risedronate but not statistically significant

Results: non-vertebral fracture Nine women in group treated with cyclical risedronate suffered non-vertebral fracture,compared with four in each of other two groups

Quality score 13/18

Comments • All patients took calcium, 1 g daily• 39 women (30%) withdrew from study: 15 (34%) from continuous risedronate group, 11 (25%) from

cyclic risedronate group and 13 (30%) from placebo group• Of women not completing study, 19 (14.4%) dropped out because of adverse events and 20 (15.2%)

because of lack of interest. Not stated from which groups they came• All women completing study had taken at least 80% of dispensed medication• Three patients in each group reported moderate to severe upper gastrointestinal adverse events• No serious adverse events were considered causally related to risedronate• Study not prospectively powered statistically to assess efficacy of risedronate on vertebral

fracture incidence• Different vertebral fracture thresholds used at two centres, so valid global fracture analysis could not

be performed• All non-vertebral fractures related to falls• Bioavailabity of risedronate may have been impaired by giving it as gelatin capsules, allowing non-dairy

fluids in period 1–2 hours before and after capsule intake, and allowing it to be taken 2 hours after meal


145



Study FIT, 199698 (women with pre-existing vertebral fractures)

Setting USA

Date of intervention May 1992–February 1996a,b



Study population Healthy postmenopausal women with established osteoporosis (at least one existing vertebral fracture)

Recruitment procedure used Population-based: each of 11 clinical centres involved developed local plan using methods such as direct mailings, media advertising, telephone solicitation, group meetings

a


Length of study Mean of 2.9 years

Main intervention/s Alendronate sodium

Primary outcome measures Incidence of new vertebral fractures

Secondary outcome measures Incidence of non-pathological clinical fractures (including both non-vertebral and symptomatic vertebral fractures)

a

Change in heighta

Changes in BMD of hip, spine, radius and total bodya

Changes in biochemical markers of bone metabolisma

Definition of incident 20% or greater decrease in anterior, middle or posterior height between baseline and endvertebral fracture of study. Additionally, for any vertebra deformed at baseline, minimum absolute change of 4 mma

Results: vertebral fracture RR of radiographic fracture in treatment group compared with control group was 0.53 (95% CI,0.41 to 0.68).This consistent regardless of age, BMD, number of pre-existing fractures or history of postmenopausal fractureb

Relative hazard of clinically apparent vertebral fractures was 0.45 (0.27–0.72)

Results: non-vertebral fracture Relative hazard of any clinical fracture in treatment group compared with control group 0.72 (95% CI,0.58 to 0.90). RR of any non-vertebral fracture 0.80 (0.63 to 1.01); hip 0.49 (0.23 to 0.99); wrist 0.52 (0.31 to 0.87); other 0.99 (0.75 to 1.31)

Quality score 18/18

Comments • 97% of participants identified themselves as Caucasian, 1% Asian, 1% African–American• All patients with estimated calcium intake at baseline of less than 1000 mg daily (81.2% of treatment

and 83.4% of placebo group) were given elemental calcium, 500 mg, + vitamin D, 250 IU, daily• Non-steroidal anti-inflammatory drugs taken for month or longer during study by about 75%

of women in both groups• Analysis undertaken on ITT basis; total number of withdrawals not given• Follow-up radiographs obtained for 1946 patients, 98% of those surviving at study close-out, of

which 1916 were obtained as part of close-out visit; for other 30 participants radiographs taken at 24 months used

• Compliance considered good – by final visit, 89% of surviving treatment group and 87% of surviving placebo group still taking medication and, of these, 96% in each group had taken at least 75% of pills since last clinic visit; however, this may be due in part to recruitment methods used

• 7.6% of women in treatment group and 9.6% in placebo group permanently discontinued study medication because of adverse experiences

• Adverse experiences resulting in hospital admission significantly less common in treatment than placebo group (24.5% versus 29.9%, p = 0.009). However, difference reduced when admission for fractures excluded (18.2% versus 20.7%, p = 0.17)

• Upper gastrointestinal problems experienced by 41.3% of women in treatment group and 40.0% in control group (p = 0.67). Rate of events did not increase after dose increased to 10 mg

• Women in treatment group had significantly fewer days in bed due to back pain than in placebo group (mean 1.9 versus 5.1 days over 3-year period, p = 0.001), and fewer days of limited activity because of such pain (mean 61.8 versus 73.2 days, p = 0.04)c

• Subgroup analysis indicated that treatment with alendronate effective even in those women at highest risk of fracture because of advanced age or severe osteoporosisb

• Authors noted that results may not be applicable to women living in institutions or in poor health

Additional information from: a Black, et al. 1993; b Ensrud, et al. 1997; c Nevitt, et al. 2000

Appendix 4

146


Study FIT, 199899 (women without pre-existing vertebral fractures)

Setting USA

Date of intervention May 1992–May 1997a



Study population Healthy postmenopausal women with osteopaenia (femoral neck BMD 0.68 g/cm2) but no vertebral fractures

Recruitment procedure used Population-based.a Each of 11 clinical centres involved developed local plan using methods such as direct mailings, media advertising, telephone solicitation, group meetings


Length of study Mean of 4.2 years


Primary outcome measures Incidence of non-pathological non-traumatic clinical fractures (including both non-vertebral and symptomatic vertebral fractures)

Secondary outcome measures Incidence of new vertebral fracturesa

Change in heighta

Changes in BMD of hip, spine, radius and total bodya

Changes in biochemical markers of bone metabolisma

Definition of incident 20% or greater decrease in anterior, middle or posterior height between baseline and end of study.a

vertebral fracture Additionally, for any vertebra that was deformed at baseline, a minimum absolute change of 4 mm

Results: vertebral fracture RR of radiographic fracture in treatment group compared with control group 0.56 (95% CI, 0.39 to 0.80);reduction significant in women whose initial T-score was –2.5 or less (RR = 0.50, 95% CI, 0.31 to 0.82) but not in those with initial T scores greater than –2.5

Results: non-vertebral fracture Relative hazard of any clinical fracture in treatment compared with control group 0.86, but reduction not significant (95% CI, 0.73 to 1.01). Relative hazard of any non-vertebral fracture 0.88 (0.74 to 1.04); hip 0.79 (0.43 to 1.44); wrist 1.19 (0.87 to 1.64); other 0.79 (0.65 to 0.96).Alendronate significantly reduced risk of clinical fractures in women with initial T-score of –2.5 or less (relative hazard 0.64; 95% CI, 0.50 to 0.82) but not in those with T-score greater than –2.5 (relative hazard 1.08; 95% CI, 0.87 to 1.35)

Quality score 18/18

Comments • 97% of participants were white• BMD cut-off chosen because considered to correspond to 2 SDs below mean for normal young adult

white women but subsequently found to correspond to 1.6 SD below mean; consequently, about one-third of women in trial actually had higher BMD than intended

• All patients with estimated calcium intake at baseline of less than 1000 mg/day (82% in each group) given elemental calcium, 500 mg, + vitamin D, 250 IU daily

• Although women taking oestrogen in preceding 6 months excluded from study, 9.2% of women in treatment group and 11.1% in placebo group took oestrogen at some time during study

• Analysis undertaken on ITT basis; number of withdrawals not given• By final visit, 81.3% of surviving treatment group and 82.5% of surviving placebo group were still taking

medication and, of these, 96% from each group had taken at least 75% of pills since last clinic visit;however, this may be due in part to recruitment methods used

• 34 participants (12 in treatment and 22 in placebo group) stopped taking study medication because their rate of bone loss exceeded predetermined limits

• Adverse events not significantly different between groups: 9.9% of women in treatment and 10.2% in placebo group permanently discontinued study medication because of adverse experiences; 29.1% of women in treatment and 26.9% in placebo group suffered adverse experiences resulting in hospital admission. Upper gastrointestinal problems experienced by 47.5% of women in treatment group and 47.2% in control group

• Follow-up radiographs obtained for 4134 women, 95% of those surviving at study close-out.

a Additional information from Black DM, et al. 1993


147



Study Harris, 1999100

Setting USA

Date of intervention December 1993–January 1998

Source of funding Procter & Gamble PharmaceuticalsHoechst Marion Roussel

Design Multicentre, double-masked, placebo-controlled, randomised

Study population Ambulatory postmenopausal women with established osteoporosis

Recruitment procedure used Variety of study centres



Main intervention/s Oral risedronate

Primary outcome measures Incidence of new vertebral fracturesIncidence of radiographically confirmed non-vertebral fracturesChanges from baseline BMD

Definition of incident New fracture defined as loss of height of at least 15% in anterior, posterior or middle height in vertebravertebral fracture that was normal at baseline, or semi-quantitatively as increase in grade from 0 to 1, 2 or 3.Worsening

fracture defined as change of 4 mm or more in vertebral height since previous radiograph, or change of grade in previously fractured vertebra

Results: vertebral fracture Cumulative incidence of vertebral fracture reduced by 41% in group taking risedronate, 5 mg, compared with placebo (95% CI, 18 to 58%; p = 0.003)

Results: non-vertebral fracture Cumulative incidence of non-vertebral fracture reduced by 39% in group taking risedronate, 5 mg,compared with placebo (95% CI, 6 to 61%; p = 0.02)

Quality score 18/18

Comments • All patients took oral calcium, 1000 mg daily• 1500 women (61%) withdrew from study; however, this figure is inflated because one arm, that taking

risedronate, 2.5 mg daily, discontinued after 1 year. Risedronate, 5 mg daily, group: 324 women (39.5%) withdrew; placebo group: 365 (344.5%) withdrew; risedronate, 2.5 mg daily, group: 163 (20.0%) withdrew before discontinued

• 86% of those who experienced vertebral fracture had at least one ‘new’ fracture (i.e. fracture of previously normal vertebra)

• Overall incidence of adverse events similar across treatment groups, as was incidence of both serious and drug-related adverse events

• Most common adverse events associated with withdrawal related to digestive system: placebo group,56 withdrawals (42%); risedronate, 5 mg daily, group, 49 (36%)

• Most upper gastrointestinal adverse events were mild to moderate in severity;incidence similar in placebo and risedronate, 5 mg daily, groups

• Patients not excluded from study on basis of history of, or ongoing, gastrointestinal disorders• No significant biochemical changes in renal, hepatic or haematological parameters were observed in

any group• Among those who withdrew, incident vertebral fractures occurred in a higher percentage of placebo

group than in risedronate, 5 mg daily, group; this may have reduced apparent treatment effect

Appendix 4

148


Study Liberman, 1995101

Setting Multinational (North America, Europe, South America, Mexico, Israel,Australia, New Zealand)




Study population Healthy postmenopausal women with osteoporosis lumbar T-score < –2.5) but no vertebral fractures





Primary outcome measures Effect on BMD at lumbar spineEffect on calcium-regulating hormonesa,b

Effect on biochemical indices of bone turnovera,b

Safety and tolerability of daily oral alendronate

Secondary outcome measures Effect on BMD at other sitesa,b

Incidence of vertebral fracturesProgression of vertebral deformitiesHeight lossSymptomatic non-vertebral fractures

Definition of incident Reduction of at least 20%, with absolute decrease of at least 4 mm, in height of any vertebral bodyvertebral fracture between baseline and follow-up

Results: vertebral fracture Treatment associated with reductions in incidence of vertebral fractures (RR 0.52; 95% CI, 0.28 to 0.95).Decreased risk still seen when stratified by age (under 65 years or 65 years and older) or the presence or absence of previous vertebral fracture

Results: non-vertebral fracture Treatment associated with trend towards reduction in incidence of fractures at non-vertebral sites (estimated risk 0.79; 95% CI, 0.52 to 1.22)

Quality score 12/18

Comments • 87.4% of patients were white, 0.4% black, 12.2% other races• Both contributory trials excluded women with history of osteoporotic fracture of proximal femur

and/or of more than one fracture of lumbar spine (this latter being to ensure that at least three vertebrae from L1–L4 were evaluable)

• Both contributory trials had three treatment groups (oral alendronate, 5 mg, 10 mg, and 20 mg reduced to 5 mg for last year, daily). Intended from outset that fracture data would be pooled as it was anticipated that numbers would otherwise not be large enough to allow detection of significant effect

• As continuous therapy with oral alendronate, 10 mg daily, produced greater decrease in incidence of vertebral fractures than other doses, pooling may have underestimated its efficacy in preventing fractures

• All patients received elemental calcium, 500 mg daily• Baseline characteristics only given for 881 women included in analysis of vertebral fractures; no

information given regarding comparability of all groups at entry• Analysis undertaken on ITT basis• Of 162 women (16.3%) who withdrew from trial, 97 (16.2%) were from treatment groups and 65

(16.4%) from placebo group. Of those in placebo group, 6.0% withdrew owing to clinical adverse events,compared with 5.4% of those taking oral alendronate, 5 mg daily, 4.1% taking 10 mg daily, and 8.0% taking 20/5 mg daily

• All four groups had similar rates of adverse upper gastrointestinal events, leading to withdrawal in placebo group of 2.0%, 3.5% in oral alendronate, 5 mg daily, 1.0% in 10 mg daily and 2.0% in 20/5 mg daily groups

• No evidence of an increased incidence of serious or severe adverse oesophageal effects seen in treatment compared with placebo groups.As severe oesophagitis has been associated with alendronate use, authors suggested results due primarily to fact that participants had regular follow-up visits with frequent reinforcement of dosing instructions, but also recognised that trial participants in general have fewer coexisting conditions than normal patientsc

Additional information obtained from: a Devogelaer, et al., 1996; b Tucci, et al., 1996; c Liberman & Hirsch, 1996


149



Study Lindsay, 1999102

Setting USA


Source of funding Merck and Company

Design Multicentre, randomised, placebo-controlled

Study population Postmenopausal women with osteoporosis and receiving HRT (T-score at lumbar spine or femoral neck < –2 and at other site < 1.5); 57% had previous fracture

Recruitment procedure used Through 38 sites


Length of study 1 year


Primary outcome measures BMD at lumbar spine

Secondary outcome measures BMD at hip trochanter and femoral neck

Biochemical markers Adverse events including clinically apparent fractures

Definition of incident Symptomatic fractures onlyvertebral fracture

Results: vertebral fracture No symptomatic vertebral fractures identified in either group

Results: non-vertebral fracture Non-vertebral fractures more common in intervention than in control group but not statistically significant (p = 0.293)

Quality score 10/18

Comments • All patients received vitamin D, 400 IU daily• All patients were receiving ongoing HRT for at least 1 year prior to study entry. Oestrogen component

of HRT was at least lowest dose recommended by manufacturer for management of osteoporosis, or approximately equivalent to at least 0.625 mg/day conjugated equine oestrogen

• All those with intact uterus received medroxyrogesterone acetate in either cyclical or continuous low-dose regimens. Other progestin preparations not permitted

• Patients whose baseline calcium intake less than 1000 mg daily provided with supplemental calcium carbonate to bring them up to this level

• Patients stratified according to duration of previous HRT to ensure equal distribution between groups of women who received HRT for less or more than 2 years. Mean duration of HRT use approximately 10 years

• Of 34 women (7.9%) who withdrew from study, 11 (5.1%) were from intervention group and 23 (10.7%) from control group

• Five women (2.3%) withdrew from intervention group and 11 (5.1%) from control group because of adverse effects. Remainder withdrew for other reasons, primarily their own request

• Adverse effects evenly distributed between two groups. Back pain was only adverse effect that was significantly more common in intervention group

• Over 90% of women in each group were at least 90% compliant with both study drug and HRT

Study Montessori, 1997103

Setting The Netherlands

Date of intervention February 1991–

Source of funding Procter & Gamble Pharmaceuticals

Design Open-label, RCT

Study population Postmenopausal women with osteopaenia (lumbar Z-score < –1); 36% had vertebral fracture on entry

Recruitment procedure used 65 participants recruited through screening programme conducted in two general practices; remainder from hospital files or incidental referrals



Main intervention/s Intermittent cyclical oral etidronate

Primary outcome measures BMD

Secondary outcome measures Vertebral fractures

Definition of incident Reduction of 20% or more in anterior, middle or posterior height, plus reduction of 10% or more invertebral fracture area in previously unfractured vertebra

Results: vertebral fracture Although trend towards lower rate of fracture in women in etidronate group, low number of incident fractures did not allow testing for statistically significant differences

Results: non-vertebral fracture No non-vertebral fractures in either group

Quality score 15/18

Comments • Both groups received calcium, 500 mg daily, etidronate group for days 15–90 of 90-day cycle, control group throughout

• Of 16 women (20%) who withdrew from study, 5 (12.5%) were from etidronate group and 11(27.5%) from control group

• Adverse events mostly mild and evenly distributed over both groups.Two cases of cancer in control group considered unrelated to study medication

• Only one patient withdrew because of an adverse event (severe diarrhoea) almost immediately after enrolment. Patient’s group was not given

• Although an open trial, radiologists who assessed spinal radiographs were blinded to treatment status• Lumbar BMD comparable in both groups at baseline. Higher proportion of women with prevalent

vertebral fractures seen in control than in etidronate group (43.6% versus 28.2%) but not statistically significant

Appendix 4

150


Study McClung, et al., 199877

Setting USA




Study population Postmenopausal women with osteopaenia (T-score at lumbar spine < –2)




Main intervention/s Oral risedronate

Primary outcome measures BMD at lumbar spine

Secondary outcome measures BMD at femoral neck and trochanterNon-vertebral fracture

Definition of incident Not applicablevertebral fracture

Results: vertebral fracture Not applicable

Results: non-vertebral fracture Few in number and comparable between groups

Quality score 7/15

Comments • Only published in abstract form• All participants received elemental calcium, 1 g daily• Two risedronate groups, 2.5 mg and 5 mg daily, and placebo group• 38% of women withdrew from study.While this not broken down between treatment arms, it

was stated that 8% of those taking risedronate and 11% of those taking placebo withdrew because of adverse events

• Incidence of mild-to-moderate upper gastrointestinal adverse events comparable between groups


151



Study Pacifici, 1988104

Setting USA




Study population White women with osteoporosis or osteopenia (at least one non-traumatic vertebral fracture and/or evidence of spinal demineralisation)

Recruitment procedure used Women attending hospital for osteoporosis screening



Main intervention/s Cyclical potassium phosphate followed by etidronateConjugated oestrogens and medroxyprogesterone acetate

Primary outcome measures Bone mineral content

Secondary outcome measures Incident vertebral fracturesTotal vertebral height lossBiochemical measures

Definition of incident Compression fractures: loss of posterior height greater than 15% compared with mean of posteriorvertebral fracture height of nearest (above and below) intact vertebrae.Wedging and biconcave fractures: loss of anterior

and central height greater than 20% compared with posterior height of same vertebra

Results: vertebral fracture Incidence of vertebral fractures almost identical in three groups. However, total vertebral height loss in hormone-treated group significantly lower (7.5 ± 4.4%, p < 0.05) than in etidronate (13.6 ± 10.6%) and control (20.8 ± 20.2%) groups, which were not significantly different from each other


Quality score 8/15

Comments • All participants received calcium, 1000 mg daily • In all, 58 women (45%) withdrew from study: numbers said to be evenly distributed between three

groups. Reasons: financial problems, geographical relocation, loss of interest, dissatisfaction with results of treatment; numbers citing each reason not given

• Baseline characteristics not presented in relation to 35 women who dropped out during first year of study; no information regarding comparability of all groups at entry

• Significant side-effects reported only in hormone group, consisting primarily of pelvic congestion and cyclic bleeding; number of women affected not specified

Appendix 4

152


Study Pols, 1999105

Setting Europe, North America, Latin America, South Africa and China



Design Multinational, randomised, double-blind, placebo-controlled

Study population Healthy postmenopausal women with osteopaenia (lumbar T-score < –2)

Recruitment procedure used 153 centres in 34 countries





Secondary outcome measures Biochemical markers of bone turnoverClinical non-vertebral fracture

Results: vertebral fracture –

Results: non-vertebral fracture Treatment associated with significant decrease in incidence of non-vertebral fractures (relative hazard 0.53; 95% CI, 0.30 to 0.90)

Quality score 11/15

Comments • All patients received elemental calcium, 500 mg daily• Of 211 women (11.1%) who withdrew, 118 (12.4%) were from treatment and 93 (9.7%) from

placebo group• No statistically significant differences found in overall incidence of adverse effects (alendronate 67.1%,

placebo 69.7%), adverse events considered by investigator to be possibly, probably or definitely drug-related (19.1% versus 18.0%) or adverse events resulting in permanent discontinuation of study medication (6.4% versus 5.6%). Serious adverse events also equally common between groups (alendronate 6.5%, placebo 6.3%)

• No significant differences between groups in overall incidence of upper gastrointestinal adverse events (alendronate 21.3%, placebo 19.3%), or specific upper gastrointestinal adverse events such as abdominal pain, dyspepsia, nausea

• Not possible to evaluate effect of treatment on vertebral fractures as baseline spinal radiographs not obtained for comparison


153



Study Reginster, 2000106

Setting Europe and Australia


Source of funding Procter & Gamble PharmaceuticalsHoechst Marion Roussel

Design Multinational, multicentre, double-masked, placebo-controlled, randomised

Study population Postmenopausal women with established osteoporosis (at least two vertebral fractures)




Main intervention/s Risedronate

Primary outcome measures Proportion of participants with at least one incident vertebral fracture

Secondary outcome measures Non-vertebral osteoporosis-related fractures (fractures of clavicle, humerus, wrist, pelvis, hip or leg regardless of trauma)Standing heightBMD at lumbar spine, femoral neck, femoral trochanter and mid-shaft radiusMarkers of bone turnover

Definition of incident Loss of height of at least 15% in anterior, posterior or middle height in vertebra that was normalvertebral fracture at baseline, or semi-quantitatively an increase in grade from 0 to 1, 2 or 3

Results: vertebral fracture Risedronate, 2.5 mg, reduced RR of vertebral fracture to 0.50 (95% CI, 0.30 to 0.84) at 12 months.Risedronate, 5 mg, reduced RR to 0.39 (95% CI, 0.22 to 0.68) at 12 months and 0.51 (95% CI, 0.36 to 0.75) at 3 years

Results: non-vertebral fracture Risedronate, 5 mg, reduced RR to 0.67 (95% CI, 0.44 to 1.04)

Quality score 14/18

Comments • All participants received calcium, 1000 mg daily• Those with baseline 25-hydroxyvitamin D below 40 nmol/l received vitamin D, up to 500 IU

daily; 35% required this supplementation, proportion being similar across all groups• Women were not excluded from study because of previous or current gastrointestinal illness or use of

medications associated with gastrointestinal intolerance (e.g. non-steroidal anti-inflammatory drugs)• Although all women had two or more baseline vertebral fractures based on initial screening radio-

graphic assessment, 2% had none and 6% only one, based on post-study serial quantitative and qualitative assessment

• There were two risedronate groups, 2.5 mg and 5 mg daily, and a placebo group. Risedronate, 2.5 mg,group discontinued after 2 years because other data showed 5 mg dose produced more consistent effect in increasing BMD, while having safety profile similar to 2.5 mg dose

• Of 684 women (60.7%) who withdrew from study, 338 (82.4%) were from 2.5 mg arm, 156 (38.2%) from 5 mg arm and 186 (45.6%) from placebo arm. Includes those who were discontinued by protocol amendment noted above

• In 2.5 mg group, 53 women (12.9%) withdrew because of adverse events, compared with 65 (15.9%) in 5 mg group and 83 (20.3%) in placebo group. No clinically meaningful differences seen between groups in incidence of adverse events

• Tablet counts indicated that 86% of women were compliant with medication (compliance defined as taking at least 80% of medication)

Appendix 4

154


Study Reid, 1994107

Setting New Zealand



Design Small, randomised, double-blind, placebo-controlled

Study population Postmenopausal women with established osteoporosis (at least one vertebral fracture)




Main intervention/s Pamidronate


Secondary outcome measures Incidence of vertebral fractures

Definition of incident Reduction of more than 20% in anterior, middle or posterior height (with, in previously fracturedvertebral fracture vertebrae, loss of height of at least 4 mm)

Results: vertebral fracture Trend towards reduction in fracture rate in treatment group (13/100 patient years versus 24/100 patient years in placebo group) but not statistically significant (p = 0.07)


Quality score 8/15

Comments • Authors recognise that size of study not intended to allow statistically significant result in relation to vertebral fracture

• All patients took elemental calcium, 1 g daily• Of 13 women (21.3%) who withdrew from trial, five (16.1%) were from treatment group and eight

(26.7%) from placebo group• Baseline characteristics only given for 48 women who completed study; there is no information

regarding comparability of all groups at entry• There was 82% compliance in each group, as assessed by tablet counts• Only common side-effects were gastrointestinal, minor in most cases, and no less common in

placebo group


155



Study Storm, 1990108

Setting Denmark

Date of intervention Patients enrolled from October 1983 to April 1986

Source of funding Norwich Eaton Pharmaceuticals

Design Small, double-blind, placebo-controlled, randomised

Study population Postmenopausal women with established osteoporosis (at least one but no more than four atraumatic vertebral crush fractures)



Length of study 150 weeks

Main intervention/s Intermittent cyclical etidronate

Primary outcome measures Bone mineral content at lumbar spine and distal non-dominant forearmSpinal deformity indexLoss of heightRate of new vertebral fractures

Secondary outcome measures Clinically overt non-vertebral fracturesBiochemical markersBone histomorphometry

Definition of incident Reduction of at least 20% in anterior, middle or posterior height (or all three), plus reductionvertebral fracture in area of at least 10%

Results: vertebral fracture Although no significant difference between overall rate of fracture in treatment and control groups from baseline to end of study, after approximately 1 year of treatment etidronate was associated with significant decrease in rate of new vertebral fractures and stabilisation in progression of vertebral deformity

Results: non-vertebral fracture Groups did not differ in terms of numbers of patients sustaining all (spontaneous and traumatic) non-vertebral fractures.All such fractures in etidronate group occurred before week 60; in placebo group,seven occurred before/during week 60 and three after week 60

Quality score 14/18

Comments • All patients received elemental calcium, 500 mg, and vitamin D, 400 IU, daily• Dietary calcium intake not recorded; no dietary restrictions or changes implemented during study• In all, 26 women (39.4%) withdrew, 13 from each group, none because of adverse events; of five deaths

in each group, none were related to either study drugs or patient’s osteoporosis• No significant side-effects related to etidronate observed • Subgroup analysis (in terms of numbers of fractures in weeks 1–60 and weeks 60–150) not pre-planned

Appendix 4

156


Study Watts, 1990109

Setting USA




Study population Healthy postmenopausal white or Asian women with established osteoporosis (at least one but no more than four vertebral crush fractures)

Recruitment procedure used Media announcements and letters to physicians



Main intervention/s Intermittent cyclical oral etidronate, with or without phosphate

Primary outcome measures Spinal BMDIncidence of new vertebral fractures

Secondary outcome measures Incidence of non-vertebral fractures

Definition of incident Reduction of 20% or more in anterior, middle or posterior height, plus reduction of 10%vertebral fracture or more in area in previously unfractured vertebra

Results: vertebral fracture Five patients (5.1%) in etidronate-only group and three (3.1%) in etidronate–phosphate group suffered new vertebral fractures compared with seven (7.6%) in phosphate-only group and ten (11.0%) in placebo-only group. Difference between etidronate–phosphate group and placebo-only group statistically significant (p = 0.034). Effect greatest in subgroup of patients whose baseline BMD was below 50th percentile of baseline values for total study population. Of these, three patients (5.8%) in etidronate-only group and three (6.1%) in etidronate–phosphate group suffered new vertebral fractures compared with seven (17.5%) in phosphate-only group and ten (21.3%) in placebo-only group.

Results: non-vertebral fracture No apparent differences seen between treatment groups in numbers of non-vertebral fractures that could be attributed to osteoporosis

Quality score 14/18

Comments • All patients given elemental calcium, 500 mg, for days 18–91 of 91-day cycle• All patients counselled in ways to achieve dietary calcium intake of at least 700 mg daily• Exceptions to stated entry criteria were granted to nine women whose weight exceeded 80 kg and

five women whose age exceeded 75 years• 66 women (15.4%) withdrew from study. Of these, six withdrew after randomisation but before

beginning of regimen. Of remaining 60, 27 (12.8%) received etidronate and 33 (15.6%) received placebo• Baseline characteristics not given for six women who withdrew immediately after randomisation• Only seven women (1.6%) withdrew because of adverse events, one (0.9%) in phosphate-only, three

(2.9%) in etidronate-only, one (0.9%) in etidronate–phosphate and two (1.9%) in placebo-only group• Adverse effects were mild, generally infrequent and comparably distributed between treatment groups:

5–6% in all groups suffered nausea during days 1–17 (phosphate/placebo and etidronate/placebo phases of cycle); however, during days 1–3 (phosphate/placebo phase), 39% of those receiving phosphate suffered diarrhoea compared with 9% receiving placebo

• Pooling of results from etidronate-treated groups and those not receiving etidronate indicated significantly fewer etidronate-treated patients with new vertebral fractures (8 versus 17, p = 0.044)

• Pooling of results for subgroup with low BMD indicated significantly fewer etidronate-treated patients with new vertebral fractures (6 versus 17, p = 0.006)

• Pooling of treatment groups and subgroup analysis in terms of BMD not pre-planned• Combination of etidronate and phosphate resulted in no apparent additional benefit beyond that

offered by etidronate alone• After first 2 years, patients could choose to continue the original blinded treatment or take calcium

alone.a Patients who completed full 3 years, whether on blinded therapy or calcium, were eligible for inclusion in 2-year, open-label follow-up study in which all patients took intermittent cyclical etidronate.They were then re-randomised to receive intermittent cyclical therapy with either etidronate or placebo.b However, only results of original 2-year double-blinded RCT are included in this review

Additional information from: a Harris, et al., 1993; b Miller, et al., 1997


157



Study Wimalawansa, 1998110

Setting UK




Study population Postmenopausal women with established osteoporosis (spinal T-score –2 and at least one but no more than four atraumatic thoracic vertebral crush fractures)




Main intervention/s HRT plus etidronate, given separately and in combination


Secondary outcome measures Biochemical markersNon-vertebral fracturesNew vertebral fracturesHeight loss

Definition of incident Reduction of 20% or more in anterior, middle or posterior vertebral height plus reduction of 15% orvertebral fracture more in area in previously unaffected vertebra. Further deterioration in height or area of previously

affected vertebra not considered a new fracture

Results: vertebral fracture Trend to fewer vertebral fractures in treatment groups than in control group (two in patients taking HRT alone, three in patients taking etidronate alone, one in patients taking combined therapy, and five in control group). However, numbers too small for results to be statistically significant even when expressed as fractures/1000 patient years

Results: non-vertebral fracture No statistically significant difference between groups in terms of non-vertebral fractures

Quality score 14/18

Comments • All participants given elemental calcium, 1 g, and vitamin D2, 400 units (10 µg) daily• All participants given advice on lifestyle, dietetic modifications, and encouraged to walk about

2 miles daily• Of 14 participants (19.4%) who withdrew, three were from the HRT group, three from etidronate

group, four from combined therapy group and four from control group. Five withdrew as result of oestrogen-related adverse effects, two from inability to tolerate medications, five from other medical problems, one died and one was lost to follow-up.Withdrawals due to toxicity were distributed as follows: HRT, three; etidronate, one; combined therapy, two; control group, one

• In all, 23 women distributed through all groups complained of minor side-effects attributable to calcium but continued supplementation

• Six women (35%) taking etidronate alone complained of nausea; no women in any other group complained of this

• Although quality of trial was relatively good, numbers were too small to produce significant results in relation to fractures as opposed to BMD

Appendix 4

158

Vitamin D derivatives

Study Aloia, 1988117

Setting USA



Design Double-blind, placebo-controlled, randomised

Study population Healthy postmenopausal women with established osteoporosis (at least one non-traumatic vertebral compression fracture)

Recruitment procedure used Media releases and letters to physicians



Main intervention/s Oral calcitriol

Outcome measures BMDIncidence of vertebral fractureBiochemical measuresBone biopsy


Results: vertebral fracture Although trend towards lower incidence in treatment group, this not statistically significant


Quality score 8/15

Comments • All patients received vitamin D, 400 IU daily• All patients had daily calcium intake assessed at entry and were instructed in 1000 mg intake.

Dietary calcium subsequently reduced to 500 mg daily in treatment group because of persistent hypercalciuria

• Baseline characteristics only given for 27 women who completed study; no information regarding comparability of all groups at entry

• Seven women (20.6%) withdrew from study, five (29.4%) from treatment and two (11.8%) from placebo group

• None of withdrawals seemed attributable to treatment; one withdrawal in placebo group due to dizziness and nausea was considered by patient to be caused by ‘drug’

• Hypercalciuria occurred in all patients treated with calcitriol; authors considered that this could have been avoided by parenteral administration of drug

• More fractures in placebo than intervention group at baseline but not statistically significant


159


Vitamin D derivatives contd

Study Dykman, 198463

Setting USA


Source of funding Not known – probably not pharmaceutical company

Design Double-blind, placebo-controlled, randomised

Study population Ambulatory rheumatic disease patients (white and black) with glucocorticoid-induced osteopenia




Main intervention/s Oral calcitriol

Primary outcome measures Forearm bone mass

Secondary outcome measures Vertebral and non-vertebral fractures


Results: all fractures Three patients (23%) in calcitriol group and four (40%) in control group sustained fractures; not statistically significant

Quality score 9/18

Comments • All patients received calcium, 500 mg, and vitamin D, 400 IU, daily• Calcitriol dose increased from 0.25 µg daily by 0.25 µg daily every 1 or 2 months as long as urinary

calcium levels remained < 350 mg/24 hours up to maximum of 1.0 µg daily; mean dose by end of study, 0.4 µg daily

• Baseline characteristics only given for 23 men who completed study; no information regarding comparability of all groups at entry

• Seven patients (23.3%) withdrew from study, three (10%) because of non-compliance but none because of adverse events; not specified from which groups they withdrew

• Toxicity frequent – 12/13 patients in calcitriol group who completed study had at least one episode of hypercalciuria or hypercalcaemia, compared with three in calcium group – but no evidence that any patients with normal renal function sustained any long-lasting complications

Study Fuji, 199264

Setting Japan

Date of intervention 1984–90


Design Randomised, open-label, controlled

Study population Women with established osteoporosis (at least one non-traumatic vertebral fracture)

Recruitment procedure used Patients with established osteoporosis consulting medical outpatient clinic of Knob Memorial Hospital


Length of study Not specified

Main intervention/s Alfacalcidol, with and without low-dose, intermittent, elcatonin (eel calcitonin derivative)

Primary outcome measures Vertebral fractures

Secondary outcome measures –

Definition of incident 20% decrease in ratio of anterior or middle height of vertebral body (whichever is larger) andvertebral fracture posterior height of same or adjacent vertebra (whichever is larger)

Results: vertebral fracture Low-dose intermittent elcatonin failed to reduce rate of vertebral fractures.Alfacalcidol seemed to be effective and effect seemed augmented by simultaneous administration of elcatonin, but too few patients to reach definite conclusion


Quality score 11/15

Comments • Patients not given calcium supplements; their calcium intake averaged 400–500 mg daily• Starting dose of alfacalcidol was 0.75 µg daily, increased stepwise to 1.5 µg daily as long as urinary

Ca/Cr stayed below 0.4. When exceeded 0.4, alfacalcidol was temporarily discontinued and then restarted at lower dose

• No specific physical therapy prescribed but adequate exercise recommended in all patients• Two patients (6.3%) withdrew, both from alfacalcidol group (25% of that group)• Quasi-randomised – patients being allocated to groups on basis of date of first visit• Groups only approximately homogeneous in relation to age and number of vertebral deformities

at baseline• Conducted 1984–90; mean duration of interventions not specified

Appendix 4

160


Study Gallagher, 1989118

Setting USA


Source of funding Hoffman La Roche(US) National Institutes of Health

Design Two comparable small, double-blind, randomised, placebo-controlled trials

Study population Postmenopausal women with established osteoporosis (at least one non-traumatic vertebral fracture)



Length of study 1 year as placebo-controlled trial

Main intervention/s Synthetic calcitriol

Primary outcome measures Vertebral fracture rates


Definition of incident Decrease of 15% in anterior height from baselinevertebral fracture

Results: vertebral fracture Combined data from two trials show lower incidence of fracture in treatment group (15 fractures) compared with placebo group (32 fractures).Authors found this statistically significant when expressed in terms of fracture rates/1000 patient years (450 compared with 823, p = 0.023)


Quality score 9/15

Comments • Data from two similar double-blind, randomised, placebo-controlled trials reported and combined.At end of first year, all patients from placebo arm crossed over to treatment; thus only results of first,placebo-controlled year reported here

• Baseline characteristics only given for 62 women with results that could be included in analysis of fracture outcomes; no information regarding comparability of all groups at entry

• All patients allowed free calcium intake• Nine patients (12.7%) withdrew from study during first year, five (13.2%) from combined

placebo arm, four (12.1%) from combined treatment arm• One withdrawal was due to patient’s death, three because patients did not satisfy initial criteria for

definition of osteoporotic fracture; no reasons given for remaining five patients dropping out


161



Study Gallagher, 1990a 119

Setting USA


Source of funding Hoffman La Roche

Design Randomised, double-blind, placebo-controlled


Recruitment procedure used Direct referral to bone clinic



Main intervention/s Calcitriol

Primary outcome measures SafetyBMD

Secondary outcome measures Incidence of vertebral fracture

Definition of incident 15% reduction in anterior or posterior vertebral heightvertebral fracture

Results: vertebral fracture No statistically significant difference between treatment and control groups


Quality score 9/15

Comments • Dose of calcitriol increased from 0.25 µg twice daily to maximum of 1.0 µg twice daily, dose being adjusted to maintain serum calcium < 2.74 mmol/l or urine calcium < 9.96 mmol/day; mean dose after 2 years, 0.62 µg daily

• All patients took vitamin D2, 400 IU daily (incorporated in multivitamin tablet)• Dietary calcium intake estimated on entry to study and all patients instructed to adjust intake to

1000 mg daily, using calcium supplements if necessary. During course of study, calcium intake was reduced to 600 mg daily to prevent hypercalcaemia


• Although double-blind, study nurse became unblinded as serum and urine calcium levels rose in first few weeks. However, she was not involved in any technical analysis; it seems to be implied, but not stated, that outcome assessors were blinded to treatment allocation

• Ten women (20%) withdrew, seven (28%) from treatment group and three (12%) from control group; in one case (treatment group), withdrawal was due to nausea

Study Orimo, 1987120

Setting Japan



Design Open-label, randomised

Study population Women with established senile osteoporosis (decreased vertebral density and at least one crush fracture)



Length of study Mean length of treatment 1.7–2.1 years

Main intervention/s Alfacalcidol, with and without calcium, compared with calcium alone and no treatment

Primary outcome measures Vertebral crush fractures


Definition of incident Reduction of 20% or more in anterior, middle or posterior vertebral height compared with baselinevertebral fracture

Results: vertebral fracture Occurrence of new fractures reduced in patients treated with alfacalcidol + calcium compared with control group (p = < 0.01). Calcium alone not effective but enhanced effect of alfacalcidol


Quality score 10/15

Comments • Although open-label, outcome assessors were blinded to treatment status• Control group received no treatment other than analgesic agents when they complained of occasional

low back pain• No information given on planned length of study or number of withdrawals. Instead, mean length of

treatment in each group given• Significant difference between calcium and control groups in number of fractures at baseline.This may

have made calcium alone appear less effective than it actually was• Although all patients described as having decreased bone mass, no definition of decreased bone

mass provided• Results statistically significant only when expressed in terms of fractures/1000 patient years. However,

these data only provided graphically; precise figures not given

Appendix 4

162


Study Orimo, 1994121

Setting Japan




Study population Postmenopausal women with established osteoporosis (decreased bone mass; 65% had fracture of spine, femur neck or radius at entry)

Recruitment procedure used Five medical institutions



Main intervention/s Alfacalcidol

Outcome measures BMDIncidence of new vertebral fracturesIncidence of non-vertebral fractures

Definition of incident Either anterior or central height 20% less than posterior heightvertebral fracture

Results: vertebral fracture Two patients (5.3%) in treatment group and seven (16.7%) in control group suffered vertebral fractures.There was 73% reduction in vertebral fracture rate/1000 patient years (from 277 in control group to 75 in treatment group, p = 0.029)

Results: non-vertebral fracture No non-vertebral fractures in either group

Quality score 12/18

Comments • All patients took elemental calcium, 300 mg daily. No specific instructions given regarding dietary calcium intake


• Six women (7.5%) withdrew from study, four (10.5%) from treatment group and two (4.8%) from control group.All seemed to be for personal reasons with exception of one in treatment group, due to side-effects

• Compliance among those who completed study satisfactory (97.3% of treatment group and 97.5% of control group followed regimen)

• No patients with overt vitamin D deficiency were included in study• Although decreased bone mass specified as inclusion criterion, cut-off level not stated• Although presence of fractures specified as inclusion criterion, 30 patients (24/53 who had adequate

X-rays for inclusion in fracture analysis) had no fractures at baseline• Women with causes of secondary osteoporosis such as bilateral oophorectomy excluded


163


Study Shiraki, 199668

Setting Japan

Date of intervention October 1989–March 1993



Study population Women aged 60 years or more with diagnosis of definite or probable osteoporosis, 49% of whom had at least one vertebral fracture at entry




Main intervention/s Alfacalcidol

Primary outcome measures BMDVertebral and non-vertebral fractures

Definition of incident Either anterior or central vertebral height 20% less than posterior heightvertebral fracture

Results: vertebral and Combined rate of vertebral and non-vertebral fracture in treatment group was one-third that in non-vertebral fracture placebo group; not statistically significant

Quality score 13/18

Comments • All patients took elemental calcium, 300 mg daily• Analgesics permitted for severe pain• In all, 34 women (30.1%) were excluded from fracture analysis, 20 (35.1%) from treatment and 14 (25%)

from control group, because of absence of X-ray films. No reasons given for this, so not clear how many excluded because of withdrawal, for whatever reason, and how many for technical reasons relating to X-rays


Study Ott, 1989122

Setting USA


Source of funding Hoffman La Roche(US) National Institutes of Health


Study population Postmenopausal women with established osteoporosis (at least two non-traumatic vertebral compression fractures)

Recruitment procedure used Media advertisements



Main intervention/s Calcitriol

Primary outcome measures Change in bone mass

Secondary outcome measures Side-effectsEffect on bone remodellingFracture incidence

Definition of incident Loss of anterior height of more than 15% resulting in anterior/posterior ratio of less than 85%vertebral fracture

Results: vertebral fracture Treatment not associated with improvement in fracture rates

Results: non-vertebral fracture Treatment not associated with improvement in fracture rates

Quality score 12/18

Comments • Calcium intake of all patients initially 1000 mg daily, using supplements if necessary• Dose of calcitriol increased from 0.25 µg twice daily to maximum of 1.0 µg twice daily; mean dose

at 96 weeks 0.43 ± 0.03 µg/day• During study, to prevent hypercalcaemia, calcium intake and dose of calcitriol adjusted to maintain

serum calcium < 2.54 mmol/l or urine calcium < 8.75 mmol/day• Of 14 women (16.3%) who withdrew from study, eight (9.3%) were from treatment group and

six (14.0%) from control group. Majority withdrew for personal reasons• Treatment group had significantly higher serum and urine calcium values but renal function not worse

than in placebo group• No major side-effects of calcitriol identified during study• Minor symptoms (e.g. upper respiratory or urinary tract infections, back pain and indigestion) similar

in both groups

Appendix 4

164

Calcitonin

Study Adami, 199593

Setting Italy











Definition of incident Not applicable: only clinically apparent fractures recordedvertebral fracture


Quality score 9/15

Comments • Alendronate arm placebo-controlled but calcitonin arm was not because no intranasal placebo available• Blinding not possible in relation to calcitonin arm and although assessors of BMD scans blinded to

treatment allocation, no such assurance was given in relation to fracture outcomes• All patients received elemental calcium, 500 mg daily• Of 41 withdrawals (14.3%) from treatment, 15 (5.2%) were due to adverse events• Analysis on ITT basis• All treatment groups similar to placebo group with regard to both overall safety profile and upper

gastrointestinal adverse events• Quality score is low partly because of lack of information about method of randomisation but more

because of lack of information about methods used to identify and confirm fractures. However, study not designed to detect significant differences in fracture rates, and fracture data only collected as part of adverse event reporting. It is therefore perhaps unreasonable to expect such detail to be provided


165


Calcitonin contd

Study Agrawal, 198161

Setting USA


Source of funding Medical Research Service of the Veterans AdministrationUS Energy Research and Development AdministrationArmour Pharmaceutical CompanyMarion Laboratories


Study population Male veterans aged 50 years and over with established osteoporosis (at least one non-traumatic vertebral compression fracture)

Recruitment procedure used Veterans Administration medical centres



Main intervention/s Subcutaneous salmon calcitonin

Outcome measures Total body calciumBone mineral contentNew vertebral eventsBiochemical indicesPeripheral fractures

Definition of incident Development of compression or collapse in vertebra previously of normal height, or furthervertebral fracture compression or collapse in previously involved vertebra

Results: vertebral fracture Eight new vertebral events in eight patients in calcitonin treatment group who had X-rays suitable for analysis, compared with 21 in seven patients in vitamin D2 + calcium group and 21 in ten patients in vitamin D2 alone group

Results: non-vertebral fracture Although fewer fractures in calcitonin treatment group than in other groups, this not subject to statistical analysis because of instances of multiple fractures in both other groups, and it was not possible to quantitate degree of trauma involved

Quality score 9/15

Comments • Calcitonin group and control group both received vitamin D2, 1300 IU, + calcium, 100 mg, daily.Third group only received vitamin D2, 800 IU daily

• Baseline characteristics only given for 26 men who completed 1–2 years of study; no information regarding comparability of all groups at entry

• Baseline characteristics of 26 patients indicate that calcitonin group had more affected vertebrae than other groups at study entry

• Of 13 (33.3%) participants who died or withdrew, five died from causes unrelated to study; initial bone biopsy found that one had osteomalacia and another normal trabecular bone volume. Remaining six (15.4%) apparently withdrew voluntarily. No information given as to which groups they withdrew from

• In all, 22 patients completed 2 full years of observation• Only one side-effect was noted, of moderately severe nausea in calcitonin group• Authors reported mean of one new vertebral event per patient in calcitonin group compared with

2.0 in vitamin D2 + calcium group and 2.4 in group receiving vitamin D2 alone.These means seem inconsistent with other figures given

Appendix 4

166

Calcitonin contd

Study Cristallini, 1993128

Setting Italy




Study population Postmenopausal women with established osteoporosis (at least two non-traumatic vertebral fractures)




Main intervention/s Intermittent cyclical intramuscular salmon calcitonin with either potassium phosphate or sodium fluoride

Primary outcome measures BMDVertebral fractures


Results: vertebral fracture No significant difference between two intervention groups and control group. Rate of fracture in intervention groups approximately same in second as in first year of treatment


Quality score 6/15

Comments • All patients received vitamin D3, 0.25 µg, + elemental calcium, 1 g: control group received this daily and calcitriol groups on days 31–75 of 90-day cycle

• Dietary calcium intake: 800–1800 mg daily• Of 48 women (38.4%) who withdrew from study, 12 (30%) were from calcitonin/potassium phosphate

group, 12 (30%) from calcitonin/sodium fluoride group and 24 (53.3%) from control group• Five women (12.5%) withdrew from calcitonin/potassium phosphate group and four (10%) from

calcitonin/sodium fluoride group because of flushing/gastrointestinal complaints.Two more (5%) withdrew from calcitonin/sodium fluoride group because of occult faecal blood and pain. Remainder withdrew due to ‘dissatisfaction’

• Because of extent of their bone loss, after 1 year patients in control group were assigned to different treatments not investigated in this study

• Potential for confounding not clear as baseline data only provided for some patients in each group and their initial comparability therefore cannot be determined


167


Calcitonin contd

Study Ellerington, 1996129

Setting England


Source of funding Sandoz LtdHeart Disease and Diabetes Research TrustCecil Rosen Foundation

Design Double-blind, placebo-controlled, RCT

Study population Healthy postmenopausal women with osteopenia (T-score in spine or hip < –1.2)




Main intervention/s Intranasal salmon calcitonin

Primary outcome measures BMD of lumbar spine and hip

Secondary outcome measures Biochemical markers of bone metabolismIncident vertebral fracture

Definition of incident Authors used Kleerekoper’s, Melton’s and height-reduction-from-baseline methodsvertebral fracture

Results: vertebral fracture Authors stated that ‘too few events occurred in any group during the study to permit analyses of fracture data’, and hence did not provide any data relating to fractures


Quality score 10/15

Comments • Dietary calcium intake assessed at baseline and at 12 and 24 months by detailed self-completion questionnaire. No patients received any calcium supplement during course of study

• Twenty women (17%) withdrew or were excluded from study• An 11% drop-out rate seen in women who received calcitonin compared with 13% in those who

received placebo• Four women withdrew because of adverse reactions (rhinitis, taste perversion or epistaxis); it was not

stated which groups they were in• 34% of women who received calcitonin suffered minor local nasal or respiratory disorders compared

with 24% of those who received placebo. Only 7% in placebo group suffered rhinitis, compared with 23% in calcitonin group

Appendix 4

168

Calcitonin contd

Study Fujita, 199264

Setting Japan





Recruitment procedure used Patients with established osteoporosis consulting medical outpatient clinic of Kanebo Memorial Hospital






Definition of incident A 20% decrease in ratio of anterior or middle height of vertebral body (whichever is larger) andvertebral fracture posterior height of same or adjacent vertebra (whichever is larger)

Results: vertebral fracture Low-dose intermittent elcatonin failed to reduce rate of vertebral fractures.Alfacalcidol seemed effective and its effect appeared augmented by simultaneous administration of elcatonin but too few patients to reach definite conclusion


Quality score 11/15

Comments • Patients not given calcium supplements; calcium intake averaged 400–500 mg daily• Starting dose of alfacalcidol, 0.75 µg daily, increased stepwise to 1.5 µg as long as urinary Ca/Cr

stayed below 0.4.When exceeded 0.4, alfacalcidol temporarily discontinued and then restarted at lower dose

• No specific physical therapy prescribed but adequate exercise recommended in all patients• Two patients (6.3%) withdrew, both from alfacalcidol group (25% of group)• Quasi-randomised – patients allocated to groups on basis of date of first visit• Groups only approximately homogeneous in relation to age and number of vertebral deformities

at baseline• Study conducted 1984–90; mean duration of interventions not specified


169


Calcitonin contd

Study Gennari, 198565

Setting Italy




Study population Postmenopausal women with established osteoporosis (at least two non-traumatic vertebral compression fractures)




Main intervention/s Injected (intramuscular or subcutaneous) salmon calcitonin

Outcome measures Bone mineral content of lumbar spine and femoral diaphysisVertebral fracturesSerum and urinary parameters


Results: vertebral fracture Fewer new vertebral fractures in two calcitonin groups than in control group but authors admit number of cases too small for any statistically significant conclusions to be drawn


Quality score 6/15

Comments • All participants received calcium, 1 g daily• While in hospital, patients started on regular diet with calcium content of about 600 mg daily with

foods low in gelatine and cartilage, and were strongly recommended to continue this at home• Six participants (7.3%) withdrew because of adverse effects (three from daily calcitonin group,

two from group receiving calcitonin on alternate days, and one from control group)• Further 12 (14.6%) were excluded from analysis because of poor compliance, and 19 because they had

not yet completed full 12 months of treatment. Not clear to which groups these patients belonged• As analysed, each group consisted of 15 participants but not clear how many had originally been

randomised to each group• Not clear whether groups were comparable at randomisation, as only characteristics of those who

completed full year are given• No major adverse effects reported. Adverse effects leading to withdrawal were gastrointestinal

discomfort (one patient in control group), nausea and vomiting (three in calcitonin groups) and flushing and fatigue (two in calcitonin groups). Also, some calcitonin-treated patients experienced mild and transient facial flushing, and mild inflammatory reaction at injection site

• Study was not apparently placebo-controlled, and not clear to what extent participants, healthcare providers and outcome assessors blinded to treatment status

Appendix 4

170

Calcitonin contd

Study Hizmetli, 1998130

Setting Turkey

Date of intervention Recruitment, October 1994–August 1995



Study population Postmenopausal women with newly-diagnosed osteoporosis (T-score < –2.5)

Recruitment procedure used Consecutive recruitment from women referred to Physical Therapy and Rehabilitation Department,Cumhuriyet University Faculty of Medicine



Main intervention/s Intranasal salmon calcitonin

Primary outcome measures Lumbar and femoral neck BMDVertebral fracture


Definition of incident Decrease in anterior vertebral height of more than 25% of posterior heightvertebral fracture

Results: vertebral fracture Although more fractures seen in control group than in both calcitonin groups, this was not statistically significant


Quality score 9/15

Comments • All patients received elemental calcium, 1000 mg daily• Patients also received vitamin D, 50,000 IU weekly, if their 24-hour urinary calcium excretion less than

120 mg, and 400 IU daily if more than 120 mg• Twenty women (18.7%) withdrew from study: six (17.1%) from group receiving calcitonin, 50 IU; six

(14.6%) from group receiving calcitonin, 100 IU; eight (25.8%) from control group. Five women (6.6%) withdrew from calcitonin groups because of side-effects; all other withdrawals due to failure to attend


171


Calcitonin contd

Study Hodsman, 1997131

Setting Canada


Source of funding Medical Research Council of CanadaRhône-Poulenc-Rorer


Study population Postmenopausal women with established osteoporosis (at least one vertebral compression fracture)




Main intervention/s Cyclical subcutaneous salmon calcitonin

Primary outcome measures Change in BMD at lumbar spine

Secondary outcome measures Change in BMD at femoral neckIncident vertebral fracture ratesBiochemical markers of bone formation and resorptionNon-vertebral fracture

Definition of incident 20% or greater reduction in posterior, middle or anterior vertebral height compared with baselinevertebral fracture

Results: vertebral fracture More vertebral fractures identified in calcitonin group than in control group. However, because of small numbers, this not statistically significant

Results: non-vertebral fracture No non-vertebral fractures occurred in either group

Quality score 11/15

Comments • Baseline characteristics only provided for 30 patients who completed study. However, characteristics of other nine said to be not significantly different

• All patients received 800 IU equivalent of human parathyroid hormone by subcutaneous injection on days 1–28 and elemental calcium, 500 mg orally, on days 71–90 of 90-day cycle

• Nine patients (23%) withdrew from study, three because of inability to learn self-injection technique,three because of localised inflammatory reactions to injections and three because underlying cancer identified

• Inflammatory reactions were considered due to formulation vehicle used in human parathyroid hormone injections rather than the peptide itself

• None of the cancer cases of appeared to be related to treatment protocol• Unspecified number of patients receiving calcitonin suffered mild nausea and skin flushing• Authors argued that vertebral fracture rates are low given that patients had average T-score of –3 or

less and at least one vertebral fracture on entry. However, as both groups received human parathyroid hormone, they cannot demonstrate that these rates are lower than those in an untreated population

Appendix 4

172

Calcitonin contd

Study Overgaard, 199260

Setting Denmark


Source of funding Danish Medical Research Council (drugs from Sandoz)

Design Double-blind, randomised, placebo-controlled

Study population Elderly women with moderate osteoporosis (T-score at distal forearm < –2), 6% of whom had vertebral fractures at entry

Recruitment procedure used Questionnaire sent to all women aged 68–72 years in six municipalities near Glostrup Hospital



Main intervention/s Salcatonin nasal spray

Primary outcome measures Bone mineral content of distal forearm and lumbar spineRates of vertebral and peripheral fractures

Secondary outcome measures Measures of bone turnover

Definition of incident Two definitions used:vertebral fracture • Wedge deformities defined as reduction of at least 25% in anterior vertebral height compared with

posterior height, and compression fractures as reduction of at least 25% in posterior height compared with adjacent vertebrae

• Essentially same except that reduction has to be at least 20% and criteria applied following correction of height of each vertebra by adjustment factor

Results: vertebral fracture Number of patients with incident vertebral fractures in the pooled salcatonin group significantly lower than in control group (four compared with six, p < 0.01)a

Results: non-vertebral fracture Numbers too low to be statistically significant

Quality score 15/18

Comments • Three salcatonin groups, receiving 50, 100 and 200 IU daily• All participants received calcium, 500 mg daily• In all, 32 participants (15.4%) withdrew – 24 (15.4%) from pooled intervention groups and eight

(15.4%) from control group• A further 12 participants completed 2 years’ treatment but for various reasons, including non-

compliance, were not ‘valid completers’, and thus not included when results calculated• Side-effects not substantially more common in treatment groups (25–33%) than in control group (23%).

General adverse events: headache, dizziness, nausea, constipation; local adverse effects: primarily nasal secretion and sneezing, nasal dryness, nasal crusts, irritation of nasal mucosa

• Although study intended as dose–response study, with rates of vertebral and peripheral fractures as primary outcome measure, number of fractures was too small to support more than comparison of all fracture types between control and pooled calcitonin groups. Pooling was not pre-planned

Additional information from: a Overgaard & Christiansen, 1996


173


Study PROOF study78 (Stock, 1997)

Setting USA



Design Multinational, multicentre, double-blind, randomised, placebo-controlled

Study population Postmenopausal women with established osteoporosis (at least one but not more than five non-traumatic vertebral fractures and T-score at lumbar spine < –2)a


Number of patients 1255b


Main intervention/s Salmon calcitonin nasal spray (100, 200 or 400 IU daily)

Primary outcome measure Time to first new vertebral fracture

Secondary outcome measure Lumbar spine BMDNon-vertebral fracture

Definition of incident Greater than 20% and greater than 4 mm reduction in anterior, middle or posterior height of vertebral fracture vertebral bodya

Results: vertebral fracture In 100 IU group, 46 fractures; in 200 IU group, 37; in 400 IU group, 46; in placebo group, 55.c RR reduction in new vertebral fractures compared with placebo was 18%, 36% and 23% for 100, 200 and 400 IU treatment groups, respectively.b At 200 IU, reduction in calcitonin group compared with placebo group statistically significant (p = 0.020)

Results: non-vertebral fracture c No non-vertebral fractures in any group

Quality score 11/15

Comments • All patients received calcium, 1000 mg, + vitamin D, 400 IU, daily.• Of 123 patients (9.8%) said to have withdrawn from study, 44 were from 100 IU group, 30 from

200 IU group, 34 from 400 IU group and 15 from placebo group.c Reasons for withdrawal not given• All analyses were ITT• Adverse effects, including those resulting in discontinuation of study drug, no more frequent in

treatment groups than in placebo group• As yet only published in abstract form. Consequently, relatively little information provided and,

as result, quality score is low and may rise when published in full• As yet, only interim results available

Additional information from: a Ginola, 1998; b Chesnut, 199878; c Kanis & McCloskey, 1999

Calcitonin contd

Study Pontiroli, 1991132

Setting Italy




Study population Women with postmenopausal or senile osteoporosis (at least one vertebral fracture)




Main intervention/s Human calcitonin, either intranasally or intramuscularly

Primary outcome measures PainMetabolic indicesBone mineral content

Secondary outcome measures New fractures


Results: vertebral fracture None in either group

Results: non-vertebral fracture None in either group

Quality score 6/15

Comments • Pilot study with very small numbers• Three women (20%) withdrew, one (14%) from intranasal and two (25%) from intramuscular group; six

in each group treated for 2 months but only four in each group for additional 4 months• Side-effects, mainly nausea and flushing, more intense and prolonged with intramuscular than with

intranasal calcitonin• Intranasal but not intramuscular calcitonin associated with significant decrease in pain score

Appendix 4

174

Study Rico, 1995135

Setting Spain




Study population Postmenopausal women with established osteoporosis (more than one non-traumatic vertebral crush fracture)




Main intervention/s Intermittent intramuscular salmon calcitonin

Primary outcome measures Spinal deformityVertebral fracture

Secondary outcome measures Cortical bone massBiochemical markersHeight

Definition of incident Reduction of at least 20% in anterior, middle or posterior vertebral height (or all three) plusvertebral fracture approximate reduction in area of at least 10% in previously unfractured vertebra

Results: vertebral fracture Over 2 years, rate of vertebral fracture was 0.07/patient-year in intervention group and 0.45/patient-year in control group (p = < 0.001). Spinal deformity index increased by 0.04 in intervention group and by 0.13 in control group (p = < 0.001)

Results: non-vertebral fracture One patient in each group sustained non-vertebral fracture or fractures

Quality score 13/18

Comments • All participants received elemental calcium, 500 mg daily, for 10 days/month• Only four participants (5.6%) withdrew, three (8.3%) from intervention group and one (2.8%) from

control group. Only one from each group withdrew because of unspecified side-effects• Although outcome assessors blinded to treatment status, not clear that either patients or healthcare

providers blinded; there is no mention of placebo

Calcitonin contd

Study Rico, 1992134

Setting Spain


Source of funding Ministry of Education and Science


Study population Postmenopausal women with established osteoporosis (at least one non-traumatic vertebral crush fracture)




Main intervention/s Intermittent intramuscular salmon calcitonin

Primary outcome measure Vertebral fracture rate

Secondary outcome measures Spinal deformity indexHeightNon-vertebral fracture

Definition of incident Reduction of at least 20% in anterior, middle or posterior vertebral height (or all three) plus vertebral fracture approximate reduction in area of at least 10%

Results: vertebral fracture A 60% reduction in new fractures in intervention group compared with 35% increase in control group (p < 0.025). Spinal deformity index increased by 0.07 in intervention group compared with 0.12 in control group (p < 0.05), and mean height reduced by 1.0 and 1.4 cm, respectively (p < 0.05)

Results: non-vertebral fracture Two patients in treatment group suffered fractures (in one case accompanied by significant trauma) versus four in control group (one accompanied by minor trauma due to fall)

Quality score 13/18

Comments • All women received elemental calcium, 500 mg, for 10 days/month• Three patients (5%) withdrew from study, two (6.3%) from treatment and one (3.6%) from control

group; one withdrawal from treatment group due to side-effects• Vertebral fracture results presented in graphic form only• Abstract describes study as ‘retrospective randomised study’ but full text does not indicate that it is

anything other than normal RCT• Although outcome assessors blinded to treatment status, not clear that either patients or healthcare

providers blinded; there is no mention of placebo


175


Calcitonin contd

Study Ringe, 1987136

Setting Germany




Study population Patients with ‘incipient to severe signs’ of steroid-induced osteoporosis




Main intervention/s Subcutaneous salmon calcitonin

Primary outcome measure Bone mineral content

Secondary outcome measures Vertebral fractures/deformitiesPainAdverse effectsNon-vertebral fracture


Results: vertebral fracture No patients in treatment group suffered vertebral fracture, compared with three in control group. One in treatment group had worse back score (an index of deformity) at end of study compared with four in control group

Results: non-vertebral fracture One patient in treatment group suffered non-vertebral fracture, compared with two in control group

Quality score 9/18

Comments • Both groups received analgesic treatment on demand and physiotherapy• Baseline characteristics only given for 36 patients who completed study; no information regarding

comparability of both groups at entry• Two patients (5.3%) withdrew from study: one dropped out of calcitonin group because of severe

nausea; one from control group died in asthmatic crisis• Adverse effects seem to have occurred in treatment group only (perhaps not surprisingly, as this seems

to have been an open trial).Three patients suffered hot flushes and three nausea; in one case, nausea was severe enough for treatment to be discontinued


Setting Germany




Study population Men and women with primary osteoporosis (undefined)




Main intervention/s Subcutaneous salmon calcitonin, daily or on alternate days

Outcome measures PainVertebral and non-vertebral fracturesBMDBone biopsy


Results: vertebral fracture Fewer fractures in intervention groups than in control group but because of small numbers involved,difference not statistically significant

Results: non-vertebral fracture Fewer fractures in intervention groups than in control group but because of small numbers involved,difference not statistically significant

Quality score 7/18

Comments • All patients received calcium, 1 g daily• Patients continued to use analgesics (mostly non-steroidal anti-inflammatory drugs) and physiotherapy

according to individual need• Baseline characteristics only given for 59 patients who completed study; no information regarding

comparability of all groups at entry• Of patients who completed study, 32% were men• Eight patients (11.9%) withdrew from study, four because of insufficient cooperation and four because

of side-effects (nausea or flushing after injection). Not specified which groups they withdrew from• Adverse events generally mild, commonest in all three groups being nausea• Calcitonin groups enjoyed significant reduction in pain

Appendix 4

176

Calcium

Study Gutteridge, 199380

Setting Australia


Source of funding Australian National Health and Medical Research CouncilMedical Research Fund of Western AustraliaOrthopaedic Research and Education Committee of the University of Western AustraliaSir Charles Gairdner Hospital Research Fund

Design Open-label RCT

Study population Postmenopausal women with established osteoporosis (at least one vertebral fracture following minor trauma)



Length of study 36 months maximum

Main intervention/s Intermittent cyclic enteric-coated sodium fluoride

Outcome measures BMDVertebral fractures

Definition of incident Not givenvertebral fracture

Results: vertebral fracture No statistically significant difference between treatment and control groups.Although authors claimed that results, excluding first year of treatment, suggest RR = 0.23 for fluoride group compared with calcium + vitamin D2 group, this was not statistically significant; moreover, the untreated group were not included in this comparison


Quality score 5/15

Comments • Patients in fluoride group and calcium + vitamin D2 group all received calcium, 1 g daily, + vitamin D2,0.5 mg (20,000 IU) weekly.There was also untreated control group

• Patients received treatment for three or four cycles (27 or 36 months) depending on vertebral radiographic and densitometric response

• Results come from pilot study preceding larger trial (Gutteridge, 199679), for which recruitment still ongoing when pilot study results published

• No information given about comparability of treatment and control groups at entry• No information given about withdrawals


177


Calcium contd

Study Recker, 1996142

Setting USA

Date of intervention Recruitment November 1987–January 1990

Source of funding National Dairy Funding and Research BoardNational Institutes of HealthLederle Laboratories


Study population Healthy elderly women with low self-chosen calcium intakes and prevalent vertebral fractures

Recruitment procedure used Volunteers from 55 study sites (mostly government-sponsored meal sites for the elderly)


Length of study Mean of 4.3 ± 1.1 years

Main intervention/s Oral calcium

Primary outcome measures Incidence of vertebral fracturesForearm bone mass changes

Definition of incident Greater than 20% reduction in anterior or posterior height relative to baseline. Positive and negative callsvertebral fracture of incident fracture by algorithm judged against clinician’s assessment, which was taken as standard

Results: vertebral fracture Calcium supplementation significantly reduced rate of incident fractures in patients with pre-existingvertebral fractures (28.3% of patients in treatment group suffered fractures, compared with 51.2% in placebo group, p = 0.023)


Quality score 12/15

Comments • Results drawn from larger study of spine antifracture and bone-sparing efficacy of calcium in elderly women of European ancestry with low self-chosen calcium intakes, with and without pre-existing vertebral fractures.Although participants not selected on basis of low BMD, study was designed to evaluate vertebral fracture in two groups: women with and without prevalent vertebral fractures on entry

• For logistical reasons, necessary to randomise patients to treatment without reference to their prevalent fracture status. Despite this, when broken down into fracture and non-fracture groups for analysis, subgroups were found to be similar in age and customary calcium intake, and prevalent fracture groups were also comparable in terms of baseline bone mineral content

• Only data on those women with pre-existing vertebral fractures are examined here• Of 251 women who originally agreed to participate in both arms of study, 54 withdrew so early that

they were regarded as having declined to participate; 18 withdrew later (12 deaths, six moved out of area).Withdrawal data not broken down by arm of study

• Median compliance overall 64%

Appendix 4

178

Oestrogen

Study Lufkin, 1992178

Setting USA


Source of funding Ciba-Geigy Corporation


Study population Postmenopausal women with established osteoporosis (low BMD and at least one vertebral fracture)

Recruitment procedure used Clinic attenders



Main intervention/s Cyclical transdermal oestradiol and oral medroxyprogesterone acetate

Outcome measures Bone turnover assessed by biochemical markers and iliac bone histomorphometryBMDIncidence of vertebral fracture

Definition of incident Decrease of more than 15% in anterior, middle or posterior vertebral height relative to baselinevertebral fracture

Results: vertebral fracture 12 women (35%) in placebo group suffered new fractures, compared with 7 (21%) in oestrogen group


Quality score 11/15

Comments • At recruitment, 38 women were receiving oestrogen, calcium supplements or vitamin D; these treatments discontinued 3 months before start of study in case of calcium, and 6 months before for oestrogen and vitamin D.Another 17 women had discontinued oestrogen use 4 or more years before study

• Women whose calcium intake estimated at less than 800 mg daily were instructed to maintain a diet providing that amount

• Eight women (10.7%) withdrew from study, 3 (8.3%) from treatment and 5 (12.8%) from placebo group• Two women withdrew from treatment group and one from placebo group because of skin reaction

to patches• Subgroup analysis showed that effect of treatment was at least as marked in older as in younger women• Previous HRT did not affect results in relation to lumbar spine BMD and vertebral fracture rate• Study objective was to evaluate effectiveness of transdermal oestrogen; however, medroxyprogesterone

acetate, used to prevent endometrial hyperplasia rather than for any effect in relation to osteoporosis,may have had additive effect to oestrogen in relation to osteoporosis


179


Oestrogen contd


Setting USA











Definition of incident Compression fractures: loss of posterior height greater than 15% compared with mean of posteriorvertebral fracture height of nearest (above and below) intact vertebrae.Wedging and biconcave fractures: a loss of anterior

and central height greater than 20% compared with posterior height of same vertebra

Results: vertebral fracture Incidence of vertebral fractures almost identical in all three groups. However, total vertebral height loss in hormone-treated group significantly lower (7.5 ± 4.4%, p < 0.05) than in etidronate (13.6 ± 10.6%) and control (20.8 ± 20.2%) groups, which not significantly different from each other


Quality score 8/15

Comments • All patients received calcium, 1000 mg daily• 58 women (45%) withdrew from study. Numbers of withdrawals said to be evenly distributed between

three groups. Reasons given: financial problems, geographical relocation, loss of interest and dissatisfaction with results of treatment; however, numbers citing each reason not given

• Baseline characteristics not presented in relation to 35 women who dropped out during first year of study; no information regarding comparability of groups at entry

• Significant side-effects said to occur only in hormone group, consisting primarily of pelvic congestion and cyclic bleeding; number of women affected not specified

Appendix 4

180

Oestrogen contd


Setting UK








Main intervention/s HRT plus etidronate, given separately and in combinationPrimary outcome measuresBMD



affected vertebra not considered new fracture

Results: vertebral fracture Trend to fewer vertebral fractures in treatment groups compared with control group (two in patients taking HRT alone, three in those taking etidronate alone, one in patients taking combined therapy and five in control group). However, numbers too small for results to be statistically significant even when expressed as fractures/1000 patient years

Results: non-vertebral fracture No statistically significant difference between groups in terms of non-vertebral fracture

Quality score 14/18

Comments • All participants received elemental calcium, 1 g, and vitamin D2, 400 units (10 µg), daily• All participants received advice on lifestyle and dietetic modifications, and encouraged to walk about

2 miles/day• Of 14 participants (19.4%) who withdrew, three were from HRT group, three from etidronate group,

four from combined therapy group and four from control group. Five withdrew as result of oestrogen-related adverse effects, two from inability to tolerate medications, five from other medical problems,one died and one lost to follow-up.Withdrawals due to toxicity: HRT group 3, etidronate group 1,combined therapy group 2, control group 1

• In all, 23 women from all groups complained of minor side-effects attributable to calcium, but continued supplementation

• Six women (35%) taking etidronate alone complained of nausea; no women from other groups complained of this

• Although quality of trial relatively good, numbers too small to produce significant results in relation to fractures as opposed to BMD


181


Oestrogen contd

Study Zarcone, 1997144

Setting Italy

Date of intervention January 1991–May 1996



Study population Postmenopausal women with osteoporosis (maximum BMD at lumbar spine 0.88 g/cm2)




Main intervention/s Conjugated equine oestrogen in association with progestogen

Primary outcome measures BMD – lumbar spineVertebral fractures

Secondary outcome measures Biochemical markers


Results: vertebral fracture 16.7% of women in placebo group suffered new vertebral fractures during study compared with 10% of those on oestrogen, 0.15 or 0.3 mg daily, and 3.3% on 0.625 mg daily

Results: non-vertebral fracture Brief reference made to treatment being associated with ‘a notable reduction’ in fractured neck of femur but no figures given

Quality score 7/18

Comments • Calcium, 500 mg daily, given to women whose daily intake less than 1200–1500 mg• No demographic details given and comparability of groups not discussed• 12 women (9.1%) withdrew; not specified from which groups they came and only one reason given

for all – ‘the common prejudices which associate the use of oestrogen replacement therapy with health risks’

Appendix 4

182

Oestrogen-like molecules

Study Maugeri, 1994157

Setting Italy

Date of intervention June 1990–November 1993



Study population Women over 65 with established osteoporosis (at least one vertebral fracture, and T-score at distal radius of < –2)




Main intervention/s Oral ipriflavonePrimary outcome measuresBMDBone metabolism marker parameters

Secondary outcome measures PainIncidence of vertebral fractures


Results: vertebral fracture Occurrence of new vertebral fractures described as minimal in treatment group and much more frequent in control group. However, actual figures not given


Quality score 7/15

Comments • All patients received calcium, 1 g daily• Not clear how many patients in each group initially• 16 women (16%) withdrew from study: three because of gastrointestinal side-effects (two in treatment

and one in control group); remainder due to protocol violations or non-compliance (6) or failure to attend for examination at 6 months (7)

• Nine women (9%) suffered side-effects: six in treatment group and three in control group. Most common side-effect was mild gastralgia during first 6 months of treatment, suffered by four women in treatment group and two in control group; it disappeared spontaneously without suspension of treatment or reduction of dose

• No significant alterations in laboratory blood parameters, including serum lipids• Treatment group showed considerable improvement in pain scores and significant reduction in

analgesic consumption• No evidence given that effect of treatment on either pain or vertebral fracture reached

statistical significance


183


Oestrogen-like molecules contd

Study Passeri, 1992158

Setting Italy




Study population Elderly women with established osteoporosis (at least one vertebral fracture, and T-score at distal radius of < –2)

Recruitment procedure used Outpatient clinics



Main intervention/s Oral ipriflavone

Primary outcome measures Bone massBiomarkers



Results: vertebral fracture No women in treatment group suffered vertebral fracture, whereas one woman in control group suffered two such fractures


Quality score 11/15

Comments • All patients received calcium, 1 g daily• Not clear whether any withdrawals other than one in placebo group who suffered vertebral fractures

and withdrew because of associated back pain• Eight patients in intervention and five in control group suffered side-effects, most common being gastric

discomfort or diarrhoea; one woman in each group suffered skin rashes.All symptoms disappeared spontaneously without interrupting treatment

• Analysis appears to have been on ITT basis• Authors considered oral administration of drug to have played positive role in terms of compliance

Appendix 4

184

Oestrogen-like molecules contd

Study Passeri, 1995159

Setting Italy


Source of funding Chiese Farmaceutici SpA

Design Two centre, randomised, double-blind, placebo-controlled

Study population Elderly women with established osteoporosis (at least one vertebral fracture, and forearm Z-score of < –2)

Recruitment procedure used Patients presenting at geriatric outpatient clinics at two university hospitals for evaluation of osteoporosis



Main intervention/s Oral ipriflavone

Primary outcome measures BMDSafety

Secondary outcome measures Incidence of new vertebral fracturesIncidence of non-traumatic peripheral fractures, or peripheral fractures due to minimum trauma

Definition of incident 20% or greater reduction in anterior compared with posterior vertebral height, or in posteriorvertebral fracture height compared with height of adjacent vertebrae

Results: vertebral fracture Four women in treatment group (20% of those for whom X-rays available) suffered new vertebral fractures compared with eight in control group (40% of those for whom X-rays available)

Results: non-vertebral fracture No women in treatment group suffered non-vertebral fractures compared with one from control group

Quality score 15/18

Comments • All patients received elemental calcium, 1 g daily• 22 women (44.9%) withdrew from study: 11 (44.0%) from treatment and 11 (45.8%) from control

group. Five dropped out because of adverse reactions (three in treatment and two in control group).Of rest, five withdrew due to of intercurrent illness, two because GPs changed therapy and ten for personal reasons or loss of interest

• Most adverse reactions were gastrointestinal symptoms.As these equally distributed between groups,authors suggested at least some may be due to calcium supplement

• Authors attributed high drop-out rate to age-related difficulty involved in these patients following study protocol for long period

• All 27 completers used at least 75% dispensed medication during first year, and 93% continued to be good compliers (defined as using 75% or more of medication) for whole 2-year period (one woman in each group used 72%)

• No serious adverse events; cardiac and pulmonary function, and blood pressure remained unchanged in both groups

• Analysis of fracture outcomes undertaken on ITT basis


185


Anabolic steroids

Study Chesnut, 198396

Setting USA


Source of funding Sterling-Winthrop Research InstituteWinthrop LaboratoriesNational Institutes of Health

Design Double-blind, randomised, placebo-controlled

Study population Postmenopausal Caucasian women with vertebral osteopaenia or one or more atraumatic spinal compression fractures

Recruitment procedure used Selected from patients self-referred or referred by their physicians to study



Main intervention/s Intermittent stanozolol

Primary outcome measures Total body calciumBone massBiochemical markers

Secondary outcome measures Incidence of vertebral fractures

Definition of incident Anterior vertebral height/posterior vertebral height x 100 = < 100%, or = 100% but with vertebral fracture biconcavity of vertebral body

Results: vertebral fracture Statistically nonsignificant trend towards reduction in vertebral fractures in treatment group


Quality score 10/15

Comments • All patients instructed by dietician on maintaining intake of elemental calcium, 1000 mg daily (by diet, supplements or both), throughout study period

• Eight women (17.4%) withdrew from study, two (8.7%) from treatment and six (26.1%) from control group; reasons for withdrawals not reported

• Authors noted that number of patients was too small and observation periods too short to produce statistically significant results in relation to vertebral fracture

• Clinical side-effects noted by 57% of patients in treatment group and 30% in control group, but none severe enough to discontinue therapy

• Increased facial hair noted by 30% of patients in treatment group (compared with 9% in control group); 22% in treatment group noted hoarseness and 9% acne compared with none in control group.Ankle oedema reported by 22% in each group

• Elevated serum glutamate oxaloacetate transaminase level demonstrated on at least one occasion by 43% of patients in treatment group (compared with 0% in control group).However, mean level over 29-month period only slightly elevated and all returned to normal following discontinuation of treatment at end of study

Appendix 4

186


Setting Australia


Source of funding Australian National Health and Medical Research CouncilMedical Research Fund of Western AustraliaOrthopaedic Research and Education Committee of the University of Western AustraliaSir Charles Gairdner Hospital Research Fund


Study population Postmenopausal women with established osteoporosis (at least one vertebral fracture following minor trauma)



Length of study Maximum of 36 months

Main intervention/s Intermittent cyclic enteric-coated sodium fluoride



Results: vertebral fracture No statistically significant difference between treatment and control groups.Although authors claimed that results, excluding first year of treatment, suggested RR = 0.23 for fluoride group compared with calcium + vitamin D2 group, this not statistically significant; moreover, untreated group not included in this comparison


Quality score 5/15

Comments • Patients in fluoride group and calcium + vitamin D2 group all received calcium, 1 g daily, + vitamin D2,0.5 mg (20,000 IU) weekly. Untreated control group also

• Patients received treatment for three or four cycles (27 or 36 months) depending on vertebral radiographic and densitometric response

• Results from pilot study preceding larger trial (Gutteridge, 199679) for which recruitment still ongoing when pilot study results published

• No information given on comparability of treatment and control groups at entry• No information given about withdrawals

Fluoride

Study Brockstedt, 199673

Setting Denmark



Design Double-blind, randomised

Study population Women with osteoporosis (undefined)




Main intervention/s Sodium monofluorophosphate

Outcome measures BMDBone markersVertebral fractureAdverse effects


Results: vertebral fracture Mean number of new vertebral fractures increased equally in both groups


Quality score 5/15

Comments • Published in abstract form only• All women received calcium carbonate, 1000 mg daily • Full 3 years of study not completed by 38 women (41%). No reasons given for withdrawals,

which were not broken down between treatment groups• No statistically significant difference between groups in terms of overall incidence of adverse effects.

In continuous fluoride group, 50% and 45.7% of women suffered adverse gastrointestinal events and adverse musculoskeletal events, respectively, compared with 56.5% and 32.6%, respectively, in intermittent fluoride group


187


Fluoride contd


Setting Australia

Date of intervention 1989–a

Source of funding Australian National Health and Medical Research CouncilMedical Research Fund of Western Australia

Design Open-label, randomised, two-centre, controlled

Study population Postmenopausal women aged under 80 years with established osteoporosis (at least one vertebral fracture but no hip fracture)




Main intervention/s Intermittent cyclic enteric-coated sodium fluoride, with or without HRT



Results: vertebral fracture Interim results only available: in non-HRT groups, significantly more patients in fluoride than control groups suffered vertebral fractures (83% versus 36%, p = 0.004). Differences more marked in first 9-month cycle than in second (24% versus 0%). In HRT groups, fewer patients in fluoride than control groups suffered such frac-tures (29% versus 56%) but not statistically significant; difference most marked in first cycle (28% versus 0%)

Results: non-vertebral fracture Stress fractures suffered by eight women (18%) in fluoride groups and none in control groups,mostly (6/8) below knee

Quality score 6/18

Comments • All patients received calcium, 1 g daily, and vitamin D2, 0.25–0.5 mg (10,000–20,000 IU) weekly• Initial dose of sodium fluoride, 60 mg, could be reduced or increased (to maximum 80 mg/day)

depending on vertebral radiographic and densitometric responsea• Of 100 patients, 35 on HRT at entry continued; randomised separately from those not on HRT• Patients not on HRT received no previous treatment for osteoporosis (other than calcium)• Patients ‘randomly adaptively assigned’ to treatment• No information given about comparability of treatment and control groups at entry• Twelve women (12%) withdrew from trial, five (14%) from HRT groups, seven (11%) from non-

HRT groups; no reasons given for withdrawals

a Additional information from: Gutteridge, 199380

Study Hansson, 1987167

Setting Sweden


Source of funding Swedish Medical Research CouncilAsker’s Foundation


Study population Postmenopausal women with established idiopathic osteoporosis (at least one, and maximum of three, vertebral compression fractures sustained during minor trauma)




Main intervention/s Sodium fluoride

Primary outcome measures Spinal bone mineral content

Secondary outcome measures Vertebral fracture


Results: vertebral fracture Number of fractures that occurred in each group too small to be statistically significant


Quality score 5/15

Comments • Three of four groups received calcium, 1 g daily. Fourth group received placebo in place of calcium• Groups comparable at entry in terms of age and BMD, but no information given regarding numbers of

prevalent vertebral fractures at entry• Twelve women (12%) withdrew from study: one (4%) from fluoride, 30 mg, group, two (8%) from

fluoride, 10 mg, group, three (12%) from calcium-only group, six (24%) from placebo-only group.Reasons for withdrawals not given

Appendix 4

188

Fluoride contd

Study Kleerekoper, 1991168

Setting USA

Date of intervention 1981–

Source of funding National Institutes of Health


Study population Postmenopausal white women with established osteoporosis (at least one vertebral compression fracture or two or more non-contiguous vertebral wedge deformities)





Primary outcome measures Vertebral fracture

Secondary outcome measures Change in cortical bone massChange in heightSide-effects, including non-vertebral fractures

Definition of incident Reduction of one or more vertebral heights by 15% or more from previous valuevertebral fracture

Results: vertebral fracture In fluoride group, 74% of women suffered vertebral fracture compared with 67% in placebo group: difference not statistically significant (p = 0.50)

Results: non-vertebral fracture In the fluoride group, 13% of women suffered non-vertebral fracture compared with 3% in placebo group: difference not statistically significant (p = 0.29)

Quality score 9/18

Comments • All patients also took calcium, 1500 mg daily• All patients instructed in active physical therapy/rehabilitation programme which continued throughout

trial• In all, 40 women (47.6%) withdrew from study, 23 (50%) from intervention and 17 (44.7%) from control

group; reasons not given. However, 22 of these women (13 fluoride, nine placebo) agreed to final follow-up visit

• Analysis on ITT basis• Power of study reduced by low recruitment and high drop-out rate to 66%• Compliance defined as taking more than 75% of study medication. Of those taking medication, 62% of

fluoride and 78% of placebo group compliant with calcium (p = 0.64), and 50% of fluoride and 72% of placebo group compliant with fluoride/placebo

• Fluoride and placebo groups differed significantly in numbers who suffered osteomalacia (17% and 0%,respectively, p = 0.01) and bone lesions in lower extremities (rate/1000 person years, 936 and 302,respectively, p = 0.02)

• In fluoride group, 109 non-vertebral bone lesions identified, with 30 in placebo group; 16 patients in fluoride and one in placebo group suffered non-vertebral bone lesions associated with pain sufficient to require dose change

• Numbers of patients who suffered at least one gastrointestinal complaint (35% in fluoride, 16% in placebo group) just reached statistical significance (p = 0.05)


189


Fluoride contd

Study Meunier, 199869

Setting France and Belgium




Study population Ambulatory postmenopausal white women with established osteoporosis (at least one vertebral fracture)




Main intervention/s Fluoride (as sodium fluoride or monofluorophosphate)

Primary outcome measures Percentage of patients with at least one new vertebral fracture between T4 and L5 after 2 years

Secondary outcome measures Percentage of patients who suffered at least one new non-vertebral fracture during 2 yearsChanges in BMDBiochemical measurementsSafety and tolerance

Definition of incident Reduction of approximately 20% or more in any vertebral heightvertebral fracture

Results: vertebral fracture Fluoride salts with calcium and vitamin D2 supplementation no more successful in preventing new vertebral fractures than calcium and vitamin D2 alone

Results: non-vertebral fracture Fluoride salts with calcium and vitamin D2 supplementation no more successful in preventing new non-vertebral fractures than calcium and vitamin D2 alone

Quality score 14/18

Comments • All patients took calcium, 1 g, and vitamin D2, 800 IU, daily• Study treatment not given to 132 women (37.3%) for full 24 months.Treatment discontinuation rates

said not to differ between groups but details not given• Lower limb pain syndrome noted more frequently in combined fluoride group than in control

group (p = 0.001). In most cases (29/37), this occurred in first 12 months of treatment; it led to definitive discontinuation of treatment in 11.5% of combined fluoride group and 3.4% of control group, and in three patients (all receiving fluoride) led to short hospitalisation period

• Authors suggested that, as proportion of patients with new vertebral fractures in control group was much lower than in previous French study, calcium and vitamin D2 may have had anti-fracture effect in control group; as result, study may have had insufficient power to detect difference in fracture rates between treatment and control groups.Alternatively, they suggested duration of study may have been too short

Appendix 4

190

Fluoride contd

Study Pak, 1995170

Setting USA

Date of intervention 1986

Source of funding United States Public Health ServiceInstitutional fundsMission Pharmaceutical Company (provision of drug only)a



Recruitment procedure used Patients referred for symptomatic osteoporosis by practising physicians because of inadequate response to conventional therapy or unwillingness of physicians to care for thema



Main intervention/s Intermittent slow-release sodium fluoride

Primary outcome measures Incident vertebral fractures

Secondary outcome measures BMDSafety

Definition of incident Reduction in any vertebral height of more than 20%, + decrease in vertebral area of more than vertebral fracture 10%, from 1 year to next

Results: vertebral fracture 85.4% of patients in treatment group remained free of new fractures in previously unaffected vertebrae compared with 56.9% in placebo group (p = 0.001). No significant difference between groups in terms of recurrent fractures

Results: non-vertebral fracture No significant difference between groups

Quality score 13/18

Comments • All patients received calcium, 800 mg daily• Randomisation stratified to take account of 31 women taking oestrogen at recruitmenta• Of 34 women (30.9%) who withdrew from study, 17 (31.5%) were from treatment and 17 (30.4%)

from placebo group. Majority of withdrawals due to lack of interest, travel problems and unrelated medical problems. One withdrawal from fluoride group due to pelvic fracture; one from control group due to nausea and vomiting

• Groups did not differ significantly in frequency of side-effects• During treatment, fluoride group enjoyed significant reduction in severity and frequency of back pain

compared with before treatment, whereas control group had only statistically non-significant reduction• Compliance, assessed by pill count, was 95.2 ± 7.9% in fluoride and 94.1 ± 6.4% in placebo group

a Additional information from: Pak, et al., 1994


191


Fluoride contd

Study Reginster, 1998171

Setting Belgium


Source of funding Rotta Research Group


Study population Postmenopausal white women with moderate osteoporosis (lumbar spine T-score of –2.5), 2% of whom had vertebral fractures at entry





Primary outcome measures Number of patients with new vertebral fractures during 4-year treatment period

Secondary outcome measures BMDBiochemical markers of bone remodellingPeripheral fractures

Definition of incident Reduction of at least 20% and absolute decrease of at least 4 mm in any height of vertebral body vertebral fracture since baseline

Results: vertebral fracture In the treatment group, 2.4% of patients suffered new fractures compared with 10% in control group (p = 0.05)

Results: non-vertebral fracture Incidence of peripheral fractures similar in both groups

Quality score 15/18

Comments • All patients received elemental calcium, 1000 mg daily• Women prescribed HRT before enrolment for purposes other than bone therapy (about 10% in each

group) continued to take it• Of 78 women (39%) who withdrew from study, 38 (38%) were from treatment and 40 (40%) from

control group• Reasons for withdrawal similar in both groups in incidence, type and severity• Lower limb pain reported by 3% of treatment and 4.7% of control group• Only 36 patients not included in analysis of vertebral fracture rate, either because of early drop-out and

no possibility of radiological follow-up (30 patients) or because their radiographs could not be evaluated (six patients)

• Compliance with study medication good and identical in both groups: on average, evaluable patients in both groups took 87% of their tablets

• If patients on HRT excluded from analysis, 2.8% of treatment group and 11.6% of control group suffered new vertebral fracture

Appendix 4

192

Fluoride contd

Study Riggs, 1990172

Setting USA


Source of funding National Institutes of Health


Study population Postmenopausal white women aged 50–75 years with established osteoporosis (low BMD at lumbar spine and at least one vertebral fracture)

Recruitment procedure used Volunteers from patients at Mayo Clinic




Primary outcome measures Vertebral fracturesSide-effects, including non-vertebral fractures

Secondary outcome measures Biochemical measurementsBMD

Definition of incident Decrease of 15% in at least one of three heights within vertebra between any two examinationsvertebral fracture

Results: vertebral fracture Using 15% fracture definition,a 55.8% of women in fluoride group suffered one or more vertebral fractures during course of study, compared with 62.8% in control group, but not statistically significant (p = 0.579).Re-analysis of results using 20% and 30% definitions of fracture resulted in even less difference between groups (20% definition – 50.0% versus 52.3%, p = 0.919; 30% definition – 32.7% versus 34.9%, p = 0.737)

Results: non-vertebral fracture More women in fluoride than control group suffered non-vertebral fractures (61 versus 24; RR = 3.2;95% CI, 1.8 to 5.6; p < 0.01).While no statistically significant difference between two groups in terms of complete fractures (35 versus 22; RR = 1.9; 95% CI, 1.1 to 3.4), significantly more women in fluoride than in control group suffered incomplete fractures (26 versus 2; RR = 16.8; 95% CI, 3.9 to 71.7; p < 0.0005)

Quality score 16/18

Comments • When recruited, 153 women receiving treatment for osteoporosis (calcium, 29; calcium + vitamin D, 30;oestrogen, 22; calcium + oestrogen, 30; calcium + oestrogen + vitamin D, 31; vitamin D, 6; oestrogen + vitamin D, 5).Treatment with calcium and/or vitamin D discontinued for 3 months and with oestrogen for 6 months before start of study)

• All patients received elemental calcium, 1500 mg daily• Patients encouraged to be active but no formal exercise or rehabilitation programme• Of 67 women (33.2%) who withdrew from trial, 35 (34.7%) were from treatment and 32 (31.7%) from

placebo group.Ten said to have withdrawn because of side-effects, ten because of other developments,and 39 for personal reasons; six died and two were lost to follow-up. No information on how reasons for withdrawal related to treatment group

• Tablet counts indicated median sodium fluoride dose = 71 mg/day and median calcium dose = 1329 mg/day in fluoride and 1305 mg/day in placebo group (not significant)

• More women in fluoride group suffered one or more side-effects severe enough to require reduction in dosage (54 versus 25; RR = 3.0; 95% CI, 1.9 to 4.8; p < 0.0001)

• Gastrointestinal pain, nausea and vomiting more common in fluoride group (17 versus 7;RR = 2.88; 95% CI, 1.2 to 7.1), although gastrointestinal bleeding and anaemia were not (9 versus 9; RR = 0.87; 95% CI, 0.3 to 2.3)

• Lower extremity pain experienced by 37 women in fluoride and five in placebo group (RR = 9.85; 95% CI, 4.0 to 24.2). More severe symptoms occurred only in fluoride group and usually disappeared 4–8 weeks after treatment stopped; in eight women they recurred when drugs were resumed at lower level

• Number of women in fluoride group suffering incomplete non-vertebral fractures likely to be underestimate as only women reporting persistent pain had X-ray of painful area

• No significant difference between treatment groups in RR of vertebral fracture when broken down by year of treatment (fluoride versus control: year 1, RR = 1.47; 95% CI, 0.4 to 5.5;year 2, 0.58; 0.3 to 1.1; year 3, 0.53; 0.3 to 1.0; year 4, 0.94; 0.5 to 1.8)

• However, data reanalysedb to show that, excluding year 1 data, and taking mean of patient numbers for years 2–4, RR = 0.57, in favour of fluoride (95% CI, 0.44 to 0.70; p < 0.0001)

• Difference in vertebral fracture incidence between two groups even less apparent when women who previously received oestrogen treatment removed from analysis

Additional information from: a Melton, et al., 1998; b Gutteridge, et al., 199380


193


Fluoride contd


Setting Germany




Study population Men with early idiopathic osteoporosis (T-score < –2.5 but no significant deformity or signs of previous vertebral fractures)

Recruitment procedure used Large cohort of men presenting at clinic with preliminary diagnosis of osteoporosis (no obvious risk factors for secondary osteoporosis) for definite diagnosis and treatment advice



Main intervention/s Intermittent oral monofluorophosphate with continuous calcium supplementation

Primary outcome measures BMD of lumbar spine

Secondary outcome measures BMD of five other sitesIncidence of fractures (vertebral and non-vertebral)Combined pain-mobility score

Definition of incident Reduction of over 20% in anterior, median or posterior vertebral height in relation to baselinevertebral fracture

Results: vertebral fracture Risk of vertebral fracture significantly reduced in fluoride group (only 10% of patients in fluoride group suffered vertebral fracture compared with 40% in control group, p = 0.015)

Results: non-vertebral fracture Risk of non-vertebral fracture also reduced in fluoride group (10% of patients suffered non-vertebral fracture compared with 27% in control group); reduction not statistically significant (p = 0.0807)

Quality score 13/18

Comments • All patients took supplementary calcium. Fluoride group took 500 mg/day for 3 months of cycle during which they also took fluoride, and 1000 mg/day for remaining month, while control group took 1000 mg/day throughout

• Fourteen men (21.9%) withdrew from study, seven (21.9%) from each group• Withdrawals due to poor compliance, change of therapy by GP, or for personal reasons not related to

study medication. No withdrawals due to side-effects• Treatment produced significant improvement in relation to pain and mobility• All reported adverse events were mild to moderate, and transient in nature• Most common adverse event in treatment group, lower-limb pain syndrome, not experienced by control

group. Disappeared after therapy interrupted for 4–5 weeks• Authors suggest that osteoporotic men may respond better to fluoride than osteoporotic women

Appendix 4

194

Fluoride contd


Setting Germany




Study population Postmenopausal women with established osteoporosis (T-score < –2.5 and at least one osteoporotic vertebral fracture)

Recruitment procedure used Patients seen in outpatients clinic



Main intervention/s Intermittent or continuous monofluorophosphate with continuous calcium supplementation

Primary outcome measures BMDIncidence of vertebral fractures

Secondary outcome measures Incidence of non-vertebral fracturesCombined pain-mobility scoreHeight loss

Definition of incident Reduction of 20% or more in anterior, median or posterior vertebral height in relation vertebral fracture to baseline

Results: vertebral fracture By third year of treatment, risk of vertebral fracture significantly reduced in both fluoride groups compared with control group. Intermittent regimen showed clear trend to be more effective than continuous regimen, with significant reduction in fracture risk by year 2

Results: non-vertebral fracture Risk of non-vertebral fracture also significantly reduced in both fluoride groups in comparison to control group; intermittent regimen showed clear trend to be more effective than continuous regimen

Quality score 13/18

Comments • All patients also took calcium, 1000 mg daily• Forty women (27.6%) withdrew from study, 14 (28.6%) from intermittent

monofluorophosphate group, 14 (29.2%) from continuous and 12 (25.0%) from control group• In first 6 months, 11 women (7.6%) withdrew for personal reasons or because of protocol violations.

They were not included in either between-group comparison of initial characteristics or, because no outcome data available, in ITT analysis

• Majority of remaining withdrawals due to non-compliance or to distance from treatment centre.However, four women withdrew from fluoride groups on their own physician’s recommendation and two because of lower extremity pain syndrome; one woman withdrew from each of continuous fluoride groups because of diarrhoea possibly related to calcium intake and one from control group because of renal colic

• Results reported here relate to new fractures only; although data also provided on worsening of existing fractured vertebrae, these not related to numbers of women who suffered them

• Treatment produced significant improvement in relation to pain and mobility• Most commonly reported adverse events were lower-limb pain syndrome and gastrointestinal

complaints; former not experienced by control group and disappeared 3–4 weeks after discontinuing therapy. Gastrointestinal symptoms occurred in all three groups; considered mainly attributable to calcium supplementation


195


Fluoride contd

Study Sebert, 1995175

Setting France




Study population Men and women with severe osteopaenia (lumbar T-score < –2) but with no vertebral fractures

Recruitment procedure used Recruited by two centres




Primary outcome measures Variations in lumbar BMD


Definition of incident A 25% reduction in anterior or middle height relative to posterior vertebral height, or in vertebral fracture vertebral height relative to adjacent vertebrae

Results: vertebral fracture Two vertebral fractures in fluoride group and one in control group; no statistically significant difference between groups

Results: non-vertebral fracture None in either group

Quality score 14/18

Comments • All patients received elemental calcium, 1000 mg daily• 35 patients (37.2%) withdrew or were considered non-eligible for trial, 19 (42.2%) from treatment and

16 (32.7%) from control group.Adequate reasons not given• Authors stated that duration of study was too short and sample size too small to draw any reasonable

conclusions on effect of treatment on fracture incidence• In fluoride group, 22% of patients suffered minor gastrointestinal disorders compared with 18% in

control group; not statistically significant (p > 0.6)• No statistically significant differences between groups in terms of numbers of patients suffering

either pain in lower limbs (fluoride 11%, control group 4%, p > 0.3) or non-vertebral bone fissures (fluoride 4%, control group 0%, p > 0.2)

• No statistically significant differences between groups in terms of tolerance of treatment;considered good/very good by over 80% patients in each group (p = 0.9)

Appendix 4

196

SERMs

Study Ettinger, 1999182

Setting Multinational


Source of funding Eli Lilly

Design Randomised, double-blind, placebo controlled

Study population Postmenopausal women with osteoporosis (low BMD and/or vertebral fractures), 37% of whom had at least one vertebral fracture at entry

Recruitment procedure used Approximately 50% of patients volunteers recruited by media advertising in USA and Canada; remainder enrolled by individual study sites which may have used own institutional or other databases to identify and contact potential patients



Main intervention/s Raloxifene, 60 mg or 120 mg

Primary outcome measures Proportion of women with one or more new non-traumatic vertebral fracturesBMD

Secondary outcome measures Proportion of women with non-traumatic non-vertebral fractures (excluding pathological fractures and those involving fingers, toes and skull)Breast cancerOther adverse eventsBiochemical markers of physiologic functions and bone turnover

Definition of incident Decrease in anterior, mid- or posterior vertebral height of at least 20% and at least 4 mmvertebral fracture

Results: vertebral fracture In pooled raloxifene groups, 272 women (6.0%) had at least one new vertebral fracture compared with 231 (10.1%) in pooled control groups; difference statistically significant

Results: non-vertebral fracture Reported by 437 women (8.5%) in pooled raloxifene groups and 240 (9.3%) in pooled placebo groups; difference not statistically significant

Quality score 16/18

Comments • Patients divided into two study groups, those with and without vertebral fractures at entry, and then randomised to receive either placebo or one of two doses of raloxifene

• All women received calcium, 500 mg, + vitamin D3, 400–600 IU, daily• Participants required to discontinue study if BMD decreased by at least 7% in lumbar spine or 10% in

femoral neck at 1 year, or at least 11% and 14%, respectively, at 2 years, or if at any time during study they experienced more than two incident vertebral fractures.As women were at high risk of non-vertebral fracture, and more women removed from placebo than from intervention group for this reason, this may have decreased study’s ability to detect statistically significant result in relation to non-vertebral fractures

• By 36 months, 1804 women (23.4%) had withdrawn from study – 652 (25.3%) from placebo and 1152 (22.5%) from raloxifene groups. However, baseline and follow-up radiographs available for 6828 women (88.6%)

• In all, 144 women (1.9%) discontinued study because of multiple fractures or excessive loss of BMD,94 (3.6%) from pooled placebo and 50 (1.0%) from pooled raloxifene groups; 754 (9.8%) withdrew because of adverse events, 227 (8.8%) from pooled placebo and 527 (10.3%) from pooled raloxifene groups (p = 0.4)

• Full baseline characteristics not provided separately for all 7705 participants as randomised to all six arms of study but only for those women with evaluable radiographs at 36 months, by study group subdivided into pooled raloxifene and placebo groups. Full baseline characteristics for all participants given but subdivided only into pooled placebo and pooled raloxifene groupsa

• Only incident fractures in vertebrae that were not fractured at baseline included in results• Venous thromboembolic events were only serious adverse effects considered to be causally related to

raloxifene treatment. By 40 months, they were reported by eight (0.3%) patients in placebo, 25 (1.0%) in 60 mg and 24 (1%) in 120 mg raloxifene groups, RR = 3.1 (95% CI, 1.5 to 6.2) in combined treatment compared with placebo group

• Breast cancer less common in combined treatment than in placebo group (RR = 0.3; 95% CI, 0.2 to 0.6)• Statistically non-significant trend towards reduced rate of non-traumatic non-vertebral fractures in

pooled raloxifene compared with control group, led to study being continued for further year• 92% of women took more than 80% of study medication; no difference between groups in

compliance terms

a Additional information from: Cummings, et al., 1999


197


SERMs contd

Study Lufkin, 1998178

Setting USA


Source of funding Eli Lilly

Design Randomised, double-blind

Study population Postmenopausal women, aged between 45 and 75 years, with established osteoporosis (low BMD and at least one non-traumatic vertebral fracture)




Main intervention/s Raloxifene, 60 mg or 120 mg

Primary outcome measures BMDBiochemical markersAdverse events

Secondary outcome measures Proportion of women with one or more new non-traumatic vertebral fracturesNumber of new atraumatic non-vertebral fracturesEffects on cognitive function and mood

Definition of incident At least 15% decrease in one or more of anterior, posterior, left lateral and right lateral vertebral vertebral fracture heights compared with baseline

Results: vertebral fracture Using fracture definition of decrease of at least 15% in vertical height, no significant difference seen between raloxifene and control groups in numbers of patients suffering incident vertebral fractures: 21(44%), 20 (43%) and 18 (38%) in 60 mg, 120 mg and control groups, respectively. However, using definition of at least 30% decrease, significant dose-dependent reduction seen, with eight (17%), four (9%) and 13 (27%) patients suffering fractures in 60 mg, 120 mg and control groups, respectively (p = 0.047)

Results: non-vertebral fracture No significant difference between the groups

Quality score 11/18

Comments • Groups generally comparable at baseline except for minor but statistically significant differences in age and alcohol usage.Thus, control group, although falling between other two groups in age, had notably higher level of alcohol usage

• All women received elemental calcium, 750 mg daily, plus vitamin D supplements to bring daily intake to 800 IU

• Although 13 women (8.9%) withdrew from study, groups from which they withdrew not identified. Eight (5.4%) withdrew due to adverse events but none considered drug-related; no significant differences between groups in numbers of adverse events reported

• Study not designed to have enough statistical power to detect differences in fracture rate• No major drug-related side-effects or adverse events reported. Of minor symptoms and signs, only

arthralgia (p = 0.027) and dizziness (p = 0.024) significantly more frequent in raloxifene groups• Raloxifene significantly decreased serum low-density/high-density lipoprotein cholesterol ratio

Appendix 4

198

Exercise

Study Ebrahim, 1997187

Setting England


Source of funding Wolfson Family Trust


Study population Postmenopausal women with established osteoporosis (upper limb fracture in past 2 years)

Recruitment procedure used Letters of invitation to consecutive suitable women identified from registers of Accident & Emergency departments and orthopaedic fracture clinics of two East London hospitals



Main intervention/s Brisk walking, building to 40 minutes three times weekly

Primary outcome measure BMD

Secondary outcome measures Fall frequencyRadiographically-identified vertebral fracturesClinical fractures

Definition of incident Difference of 25% between anterior and posterior vertebral heightsvertebral fracture

Results: vertebral fracture No significant difference between treatment and control groups


Quality score 12/18

Comments • All patients received advice about general health and balanced diet• Control group performed upper limb exercises• Acceptability of intervention low – only 165 of 508 women contacted (32.5%) agreed to take part, and

only 97 (19.1%) completed 2 years. However, study took place in relatively poor inner-city population and may not be typical of other areas

• Of 165 women participated, 68 (41%) did not complete 2 years. Drop-outs said to be evenly distributed between brisk walking and control groups. Reasons for withdrawal after randomisation given as unwillingness to continue (24%), illness (6%), death (1%), exercise-related trauma (1%) and other unspecified difficulties (9%)

• Women who dropped out tended to be less physically fit than others; thus those patients may have been lost who would have benefited most from intervention

ª Compliance with walking self-reported by those who completed study as good, but could not be validated

• Brisk walking group experienced more falls than control group (an excess of 15.2 falls/100 person years over course of study)

• Following intervention, no significant differences seen between groups’ Nottingham Health Profile scores. However, brisk walking group showed greater improvements in stamina


199


Protein supplements

Study Schurch, 1998186

Setting Switzerland

Date of intervention April 1992

Source of funding Sandoz Nutrition LtdSwiss National Research Science Foundation

Design Randomised, double-blind, placebo-controlled, with 6-month post-treatment follow-up

Study population Men and women, aged over 60 years, with recent osteoporotic hip fracture

Recruitment procedure used Patients in orthopaedic ward of Geneva Hospital


Length of study 6 months with 6-month post-treatment follow-up

Main intervention/s Oral protein supplement

Primary outcome measures Insulin-like growth factor-1 levels and other biochemical dataBMD

Secondary outcome measures Vertebral deformityLength of hospital stayFunction

Definition of incident Decrease of more than 20% in anterior, middle or posterior vertebral height from baselinevertebral fracture

Results: vertebral fracture Trend to fewer vertebral fractures in treatment group was not statistically significant


Quality score 13/15

Comments • All patients received single oral 200,000 IU dose of vitamin D3

• Of 19 patients (23%) who withdrew from trial or died during 6-month intervention period, 11 (26.8%) were from treatment and eight (19.5%) from control group. Further eight (9.8%) dropped out or died during 6-month follow-up period, six (14.6%) from treatment, two (4.9%) from control group

• Treatment well tolerated; no major side-effects• Four patients (9.8%) withdrew from each group because of nausea, and two (4.9%) from treatment and

one (2.4%) from control group because of diarrhoea• Treatment group stayed in orthopaedic ward for 18.0 ± 1.4 days post surgery, and control group

16.9 ± 0.9 days, but after transfer to rehabilitation hospitals, median stay in treatment group was 21 days shorter than control group (p = 0.018)

Appendix 4

200

Vitamin K2


Setting Japan

Date of intervention Not stated

Source of funding Not stated

Design Randomised, open-label, placebo-controlled

Study population Ambulatory women with osteoporosis (lumbar BMD < 70% of young adult mean, or with one or more non-traumatic vertebral fractures and lumbar BMD < 80% of young adult mean)

Recruitment procedure used Patients with osteoporosis registered with Research Institute and Practice for Involutional Disease, Nagano, Japan



Main intervention/s Vitamin K2

Primary outcome measures Lumbar BMDVertebral fracture

Secondary outcome measures Biochemical markersNon-vertebral fracture

Definition of incident Decrease of 20% or more in anterior, middle or posterior vertebral height from baselinevertebral fracture

Results: vertebral fracture Thirteen vertebral fractures in 77 women in vitamin K2 group compared with 30 in 64 women in control group

Results: non-vertebral fracture One non-vertebral fracture in 77 women in vitamin K2 group compared with five in 64 women in control group

Quality score 12/18

Comments • Patients in both arms received elemental calcium, 150 mg daily• Patients not given any specific instructions regarding daily calcium, vitamin D or vitamin K intake, or

exercise; and prohibited from taking any other drugs that affected bone or calcium metabolism during study period

• Only 190 women had X-rays at 24 months, allowing them to be included in fracture analysis. No information why such data not available for remaining 51 women – 29 (24.2%) from vitamin K2 and 22 (18.2%) from control group

• Combined incidence of vertebral and non-vertebral fractures significantly (p = 0.0273) lower in vitamin K2 than in control group

• No information on side-effects or adverse events, if any


201


Comparisons with active treatments

Study Abellan Perez, 1995139

Setting Spain




Study population Postmenopausal women with established osteoporosis (at least one non-traumatic vertebral crush fracture)

Recruitment procedure used Through outpatient rheumatology clinics at 12 hospitals



Main intervention/s Intranasal synthetic salmon calcitonin and calcium compared with larger dose of calcium

Primary outcome measures PainNew vertebral fractures

Secondary outcome measures Limitation of movementBiochemical parameters

Definition of incident Meunier’s index of spinal deformityvertebral fracture

Results: vertebral fracture Authors claimed statistically significant (p < 0.001) reduction in rate of vertebral fracture in intervention group; however, data presented in such a way that precise extent of reduction not clear


Quality score 10/15

Comments • Both intervention and control groups received elemental calcium: intervention group received 500 mg for 14/28 days and control group received 1000 mg daily throughout

• Significant reductions in intensity of pain, limitation of action by pain, and analgesic use in intervention group

• Only one woman from intervention group (2.3%) is stated to have withdrawn, as result of arterial hypertension which might have been treatment-related. However, other participants may have withdrawn for other reasons

• Other side-effects were mild: one case of flushing and three gastrointestinal problems

Study Adami, 199593

Setting Italy











Definition of incident Not applicable: only clinically apparent fractures were recordedvertebral fracture


Quality score 9/15

Comments • Alendronate arm placebo-controlled but not calcitonin arm because intranasal placebo not available• Blinding not possible in relation to calcitonin arm and, although assessors of BMD scans blinded to

treatment allocation, no such assurance given for fracture outcomes• All patients received elemental calcium, 500 mg daily• Of 41 women (14.3%) who withdrew from treatment, 15 (5.2%) were due to adverse events• Analysis was on an ITT basis• All treatment groups similar to placebo with regard to both overall safety profile and upper

gastrointestinal adverse events• Quality score low because of lack of information, partly about method of randomisation but more

about methods used to identify and confirm fractures. However, study not designed to detect significant differences in fracture rates, and fracture data only collected as part of adverse event reporting. It is thus perhaps unreasonable to expect such detail to have been provided

Appendix 4

202

Comparisons with active treatments contd

Study Arthur, 1990190

Setting USA



Design Randomised trial

Study population Elderly postmenopausal women with radiographic evidence of osteopaenia; 40% had vertebral compression fractures at entry


Number of patients Fourteen in randomised study, plus four age-matched normal individuals


Main intervention/s Low-dose calcitriol compared with vitamin D2


Secondary outcome measures Biochemical analysesVertebral fractures


Results: vertebral fracture No-one in either group developed new compression fractures, even though each group contained two women with prevalent fractures and at high risk of new fractures

Results: non-vertebral fracture No-one in either group suffered non-vertebral fracture

Quality score 7/15

Comments • Initial dose of calcitriol was 0.25 µg/day, increased to 0.25 µg twice daily if (? blood or urine) calcium level remained at 10 mg/dl or less. By end of study, all patients in calcitriol group taking 0.25 µg twice daily

• Patients in vitamin D2 group received 50,000 units orally twice weekly• All patients received elemental calcium, 1 g daily• Patients not placed on calcium-restricted diet• Only data relating to two randomised treatment groups considered here• Three women (21.4%) withdrew: one from calcitriol group; two from vitamin D2 group.Another

woman (group unknown) excluded from analysis due to osteomalacia• Baseline characteristics only given for ten women who completed randomised study; no

information given regarding comparability of all groups at entry• Combination of calcitriol and calcium can lead to hypercalciuria, hypercalcaemia and renal failure.

Significant hypercalciuria observed in both groups and transient hypercalcaemia in one individual in vitamin D2 group. However, patients’ renal functions did not decline significantly over course of study


203



Study Birkenhager, 1992164



Source of funding Organon Nederlands BV

Design Single-blind RCT

Study population Postmenopausal women with established osteoporosis (at least one non-traumatic vertebral compression fracture) or osteopaenia (lumbar BMD > 2 SD below normal mean of age-matched controls) (86% had fractures)




Main intervention/s Intermittent intramuscular nandrolone decanoate + HRT compared with HRT alone

Primary outcome measures BMDBiochemical markers

Secondary outcome measures Vertebral fracturePhoniatrics

Definition of incident Reduction in anterior and/or mid-vertebral height of at least 20% of posterior height, orvertebral fracture reduction of at least 20% in posterior height of one or both adjacent vertebral bodies

Results: vertebral fracture Mean number of deformed vertebrae per individual did not increase significantly in either group


Quality score 11/15

Comments • Patients whose dietary history suggested that this necessary received oral calcium supplements up to total of at least 1000 mg/day

• Seven women (16.7%) withdrew from study, five because of voice problems and two because of extent of uterine bleeding. Not stated from which groups they withdrew

• Number of patients too small and follow-up too short to produce statistically significant results in relation to vertebral fracture

• Substantially higher proportion of patients in nandrolone group than in control group complained of vocal changes (p < 0.01); these confirmed by logopaedic evaluation. Not clear whether changes would prove to be reversible

• No patients complained of increased hair growth, whether facial or other• Total serum cholesterol unchanged in nandrolone but fell in control group; high density lipoprotein

cholesterol fell in nandrolone but unchanged in control group• Potential for confounding not clear as baseline data only provided for some of patients in each group;

hence, comparability at entry cannot be determined

Appendix 4

204


Study Ebeling, 199870

Setting Australia



Design Randomised, double-masked, double-placebo controlled trial

Study population Men with moderately severe idiopathic osteoporosis (at least one vertebral fragility fracture)


Number of patients Specified as both 39 and 41


Main intervention/s Calcitriol compared with calcium

Primary outcome measures Vertebral fracturesSpinal BMD

Secondary outcome measures Non-vertebral fractures


Results: vertebral fracture Significantly more patients in one arm than in other suffered vertebral fracture

Results: non-vertebral fracture Six non-vertebral fragility fractures in one arm and none in other

Quality score 8/18

Comments • Published in abstract form only• Abstract presents groups in masked form only• Five patients withdrew from study. It was not stated from which arm they withdrew

Study Falch, 1987191

Setting Norway



Design Single-blind RCT

Study population Postmenopausal women with recently sustained fracture of distal left forearm




Main intervention/s Calcitriol compared with vitamin D3

Outcome measures Bone massVertebral fracturesFractures of long bones

Definition of incident Each vertebra compared with nearest cranial vertebra. If anterior wedging or middle depression was vertebral fracture < 85% of cranial vertebra, vertebra assigned fracture score of 1; if both measurements < 85%, fracture

score was 2

Results: vertebral fracture No significant difference between treatment groups

Results: non-vertebral fracture No significant difference between treatment groups

Quality score 13/18

Comments • Patients’ allocation known to clinician in charge of study but not to those evaluating their skeletal status• Ten women (11.6%) withdrew, eight (17.0%) from calcitriol group and two (5.1%) from vitamin

D3 group, none because of drug-related adverse effects• Initial dose of calcitriol, 0.50 µg daily, halved if total serum calcium exceeded 2.65 mmol/L.

This necessary in 11 women (28%)


205




Setting Japan





Recruitment procedure used From patients with established osteoporosis consulting medical outpatient clinic of Kanebo Memorial Hospital






Definition of incident Decrease of 20% in ratio of anterior or middle height of vertebral body (whichever is larger)vertebral fracture and posterior height of same or adjacent vertebra (whichever is larger)

Results: vertebral fracture Low-dose intermittent elcatonin failed to reduce rate of vertebral fractures.Alfacalcidol appeared effective, with effect augmented by simultaneous administration of elcatonin, but too few patients to reach a definite conclusion


Quality score 11/15

Comments • Patients not given calcium supplements; calcium intake averaged 400–500 mg/day• Starting dose of alfacalcidol 0.75 µg/day.This increased stepwise to 1.5 µg as long as urinary Ca/Cr

stayed below 0.4.When exceeded 0.4, alfacalcidol temporarily discontinued, and then restarted at lower dose

• No specific physical therapy prescribed but adequate exercise recommended for all patients• Two patients (6.3%) withdrew, both from alfacalcidol group (25% of group)• Study quasi-randomised, patients being allocated to groups on basis of date of first visit• Groups only approximately homogeneous in relation to age and number of vertebral deformities

at baseline• Conducted from 1984 to 1990; mean duration of interventions not specified

Appendix 4

206



Setting Japan



Design Multicentre, double-blind placebo-controlled

Study population Patients with postmenopausal or senile osteoporosis (scoring 4 points or more on diagnostic criteria proposed by Osteoporosis Study Groups of Japanese Ministry of Health and Welfare for diagnosis of involutional osteoporosis




Main intervention/s Etidronate (compared with alfacalcidol)

Primary outcome measures Lumbar vertebral mineral density

Secondary outcome measures Incidence of vertebral fracturesBiochemical parametersPain relief

Definition of incident Not given in abstractvertebral fracture

Results: vertebral fracture Patients in two etidronate groups had fewer fractures than alfacalcidol comparison group


Quality score 6/15

Comments • Quality score very low because no translation of article available and hence data could only be extracted from abstract. Paper is long and appears detailed; it is therefore possible that quality score would increase considerably were it possible to extract full data

• Some uncertainty about comparability of different arms of study. One etidronate group appeared more severely osteoporotic than other; not specified how compared with alfacalcidol group

Study Gallagher, 1990b 74

Setting USA



Design RCT

Study population Men and women with established idiopathic spinal osteoporosis (at least one vertebral fracture)



Length of study At least 3 years

Main intervention/s Comparison of fluoride + calcium with fluoride + calcitriol

Outcome measures HyperosteoidosisTotal body calciumVertebral fractureHip fracture


Results: vertebral fracture Incidence at end of first year = 60% in each arm

Results: non-vertebral fracture Incidence during first 18 months = 6% overall

Quality score 6/18

Comments • Abstract only• Patients included 35 women and five men. No information given on groups to which randomised• No information given on baseline characteristics at entry• At time of writing, 31 patients completed 1 year of therapy, 20 2 years and 13 3 years


207



Study Geusens, 198663

Setting Belgium


Source of funding Supply of drugs only: Organon International; Organon Belgium; Leo Pharmaceutical,Belgium/Denmark; Sandoz, Basle, Switzerland

Design Double-blind, randomised, placebo-controlled trial

Study population Men and women with established osteoporosis (at least one vertebral fracture)

Recruitment procedure used Consecutive patients admitted to unit because of vertebral collapse without trauma



Main intervention/s Intermittent intramuscular nandrolone decanoate compared with alfacalcidol and with intermittent intravenous calcium infusions

Primary outcome measures Bone mineral contentCortical bone volumeVertebral fracture rateBiochemical variables

Secondary outcome measures Non-vertebral fractures

Definition of incident Clear change (> 20%) in shape of anterior, middle or posterior part of vertebravertebral fracture

Results: vertebral fracture Statistically nonsignificant trend to lower fracture rates in second year of treatment in nandrolone compared with alfacalcidol and intermittent calcium infusion groups.Vertebral fracture rate 40% lower in nandrolone than in other groups 2 years after end of treatment but, again, not statistically significant


Quality score 10/15

Comments • Of 26 patients (43%) who withdrew from study, one withdrew from each group because of ‘subjective tolerance without organic side-effects’. Remainder, not attributed to treatment groups, withdrew because of intercurrent illness or discontinuation of treatment

• Potential for confounding not clear as baseline data only provided for patients who completed study;hence, comparability of groups at entry cannot be determined. Of patients completing trial, those in calcium group significantly older than other two groups

• No serious side-effects reported or noted; no signs of virilisation or disturbance of liver function in nandrolone group

Study Guañabens, 199675

Setting Spain




Study population Women with established postmenopausal osteoporosis (mean lumbar BMD T-score –3.2 SD + at least one atraumatic vertebral fracture)




Main intervention/s Fluoride compared with cyclic etidronate

Primary outcome measures Bone massVertebral fracture

Secondary outcome measures Non-vertebral fracture


Results: vertebral fracture More patients in etidronate than in fluoride group had new vertebral fractures (26 fractures in 11 women versus nine in seven women); not statistically significant

Results: non-vertebral fracture Rate of non-vertebral fracture said to be similar in both groups; figures not given

Quality score 7/18

Comments • Abstract form only• All patients also took calcium, 1 g, ‘regularly’• Of 36 women (28.6%) who withdrew from study, 21 (35%) were from fluoride and 15 (22.7%) from

etidronate group• Four withdrawals from fluoride group due to stress fractures and eight to gastrointestinal symptoms;

no withdrawals from etidronate group attributed to either cause

Appendix 4

208

Study Leidig-Bruckner, 199776

Setting Germany



Design Open-label, randomised




Length of study Mean follow-up 4.7 ± 1.6 years

Main intervention/s Sodium fluoride alone or with either HRT or nandrolone decanoate



Results: vertebral fracture Progression of vertebral fractures not different between groups, but highly related to severity of fracture status at beginning of therapy


Quality score 7/15

Comments • Abstract form only• All patients received calcium, 1000 mg, and vitamin D, 3000 IU, daily• Nandrolone decanoate given for 2 years; implied but not stated that other interventions given for at

least 4 years• Three women (7.9%) withdrew from study; not clear from which groups they withdrew



Setting Australia

Date of intervention 1989–a

Source of funding Australian National Health and Medical Research CouncilMedical Research Fund of Western Australia

Design Open-label, randomised, two-centre, controlled

Study population Postmenopausal women aged under 80 years with established osteoporosis (at least one vertebral fracture but no hip fracture)




Main intervention/s Intermittent cyclic enteric-coated sodium fluoride, with or without HRT



Results: vertebral fracture Interim results only available. In non-HRT groups, significantly more patients in fluoride than control group suffered vertebral fractures (83% versus 36%, p = 0.004). Differences more marked in first 9-month cycle than second (24% versus 0%). In HRT groups, fewer patients in fluoride than control group suffered such fractures (29% versus 56%) but not statistically significant; difference most marked in first cycle (28% versus 0%)

Results: non-vertebral fracture Eight women (18%) in fluoride and none in control groups suffered stress fractures, mostly (6/8) below knee

Quality score 6/18

Comments • All patients received calcium, 1 g daily, + vitamin D2, 0.25–0.5 mg (10,000–20,000 IU) weekly• Initial dose of sodium fluoride, 60 mg;a could be reduced or increased (to maximum of 80 mg/day)

depending on vertebral radiographic and densitometric response• Of 100 patients, 35 on HRT at entry continued; randomised separately from those not on HRT• Patients not on HRT had not previously received treatment for osteoporosis (other than calcium)• Patients ‘randomly adaptively assigned’ to treatment• No information given on comparability of treatment and control groups at entry• Twelve women (12%) withdrew, five (14%) from HRT and seven (11%) from non-HRT groups;

no reasons given

a Additional information from: Gutteridge, 199380


209



Study Lems, 1997192


Date of intervention July 1991–

Source of funding Dutch League against RheumatismChristiansen (supplier of sodium fluoride and placebo)


Study population Men and women with established corticosteroid-induced osteoporosis (vertebral deformity,prior peripheral fracture or both)

Recruitment procedure used Hospital departments of rheumatology and clinical immunology, nephrology and pulmonology



Main intervention/s Cyclical etidronate with or without enteric-coated sodium fluoride


Secondary outcome measures Vertebral fracturesPeripheral fracturesBiochemical markers

Definition of incident Decrease of 15% in anterior, middle or posterior vertebral height relative to baseline. Clinicalvertebral fracture vertebral deformity defined as vertebral deformity causing clinical manifestations leading to

prescription of therapy (bed rest, analgesia or both)

Results: vertebral fracture Seven patients in fluoride and four in control group suffered vertebral fractures. Difference between groups not statistically significant

Results: non-vertebral fracture Eight patients in fluoride and five in control group suffered non-vertebral fractures. Difference between groups not statistically significant

Quality score 11/18

Comments • All received cyclical etidronate for 2 weeks followed by 11 etidronate-free weeks• All received supplement of elemental calcium, at least 500 mg daily; those with low dietary calcium

intake received 1000 mg• Patients with serum 25 hydroxyvitamin D concentration below 10 µg/L in winter and 15 µg/L in

summer, received vitamin D (dihydrotachysterol, 0.2 mg, on alternate days)• All took prednisone, at least 7.5 mg daily, at start of study and were expected to continue for at least

6 months• Seven women (14.9%) withdrew from study, three (13.0%) from fluoride and four (16.7%) from control

group: one from fluoride and two from control group did not wish to travel for BMD measurement;two from control group because of other severe diseases, one from fluoride group because of incomplete fractures at knee and another died of pulmonary embolism

• No patient in fluoride group reported gastrointestinal complaints• Lower extremity pain syndrome reported by only one patient taking fluoride• Lumbar spine BMD significantly lower in fluoride than control group, approximately doubling risk of

fracture in intervention group

Study Lyritis, 1994166

Setting Greece




Study population Postmenopausal women with established osteoporosis (at least one non-traumatic vertebral collapse)




Main intervention/s Intramuscular nandrolone decanoate compared with alfacalcidol

Primary outcome measures Incidence of vertebral fracturesBMDSymptoms (pain, mobility)

Secondary outcome measures Biochemical markers

Definition of incident Meunier–Vignon indexvertebral fracture

Results: vertebral fracture No recent vertebral fractures in either group


Quality score 9/15

Comments • Number of withdrawals not specified• Nandrolone improved pain and mobility significantly more than alfacalcidol

Appendix 4

210


Study Mamelle, 1988176

Setting France

Date of intervention Recruitment: November 1984–December 1985

Source of funding Merck-Clevenot LaboratoriesINSERM

Design Large, open-label RCT

Study population Men and women with established primary vertebral osteoporosis (at least one non-traumatic vertebral crush fracture)

Recruitment procedure used Patients consulting one of 94 physicians who agreed to participate



Main intervention/s Sodium fluoride, calcium and vitamin D2 compared with other regimes prescribed by French physicians


Secondary outcome measures Side-effects (including non-vertebral fractures)

Definition of incident Deformation of normal vertebra to biconcave one, a fracture plate, or wedging or collapse of vertebralvertebral fracture body

Results: vertebral fracture Of patients who could be followed-up for 24 months, 39.2% of fluoride group had one or more new vertebral fractures compared with 50.8% of control group (p < 0.05)

Results: non-vertebral fracture Of those patients who could be followed up for 24 months, 12.8% of fluoride group had one or more new non-vertebral fractures compared with 12.5% of control group

Quality score 11/18

Comments • Approximately 10% of patients in both groups were male• Fluoride group received elemental calcium, 1 g, + vitamin D2, 800 IU, daily• Control group received following treatments:

– calcium, 1 g, + vitamin D2, 800 IU, daily (n = 95)– calcitonin, 500 U daily, for 5 days/3 weeks, + phosphorus, 1–1.5 g daily (n = 85)– calcitonin + calcium (n = 12)– calcium + phosphorus (n = 17)– phosphorus + sodium etidronate, taken for 5 days/3 weeks (n = 2)

• In all, 240 patients (51.5%) did not continue treatment for full 24 months, although some were followed-up for full period: 128 (49.8%) from fluoride and 112 (53.6%) from control group. In fluoride group, 24 patients (9%) discontinued treatment because of intolerance, together with 17 (8%) from control group; remainder withdrew because of death, intercurrent disease, aggravation of vertebral osteoporosis, or personal reasons unrelated to treatment, or were lost to follow-up

• In all, 316 patients (68%) were followed-up for full 24 months• No significant difference between two groups in relation to osteo-articular pain, non-vertebral fractures,

gastrointestinal disorders and other side-effects, although pain in lower limbs more often located in ankle and foot in fluoride group than control group (15% versus 5%, p < 0.01). However, full details of side-effects only given for patients who completed 24 months of treatment; these patients may have suffered lower rate of side-effects than those who withdrew from study. In total sample, side-effect data presented only as RRs and, when adjusted for previous osteo-articular symptoms, indicated that for osteo-articular pain, RR = 0.73 in control versus fluoride group – not statistically significant (p = 0.07);however, for ankle and foot pain, RR = 0.34 in control group – statistically significant (p = 0.01)


211




Setting USA











Definition of incident Compression fractures: loss of posterior height greater than 15% compared with mean ofvertebral fracture posterior height of nearest (above and below) intact vertebrae

Wedging and biconcave fractures: loss of anterior and central height greater than 20% compared with posterior height of same vertebra

Results: vertebral fracture Incidence of vertebral fractures almost identical in all three groups. However, total vertebral height loss in hormone-treated group significantly lower (7.5 ± 4.4%, p < 0.05) than in etidronate (13.6 ± 10.6%) and control (20.8 ± 20.2%) groups, which not significantly different from each other


Quality score 8/15

Comments • All patients received calcium, 1000 mg daily • In all, 58 women (45%) withdrew from study. Numbers of withdrawals said to be evenly distributed

between three groups. Reasons: financial problems, geographical relocation, loss of interest and dissatisfaction with results of treatment; however, numbers citing reasons not given

• Baseline characteristics not presented in relation to 35 women who dropped out during first year of study; no information on comparability of all groups at entry

• Significant side-effects said to occur only in hormone group; consisted primarily of pelvic congestion and cyclic bleeding; number affected not specified

Appendix 4

212


Study Pak, 1989193

Setting USA



Design Open-label, randomised trial

Study population Men and women with osteoporosis or osteopaenia (fractures, low BMD or radiological evidence of osteopaenia) (98% had vertebral or peripheral fractures)



Length of study Maximum of 5.5 years

Main intervention/s Intermittent slow release sodium fluoride, with or without preceding high dose of 1,25-dihydroxyvitamin D

Outcome measures BMDVertebral fracturesNon-vertebral fracturesBone biopsy

Definition of incident Reduction of more than 15% in vertebral heightsvertebral fracture

Results: vertebral fracture Vertebral fracture rates significantly reduced in both arms of study, especially after 1 year of treatment,compared with same patients prior to intervention. Fracture rates lower in group not receiving 1,25-dihydroxyvitamin D than in group receiving it, although former group had history of more fractures than latter

Results: non-vertebral fracture One hip fracture in 1,25-dihydroxyvitamin D group and two other fractures in group not receiving it

Quality score 15/18

Comments • Each group contained three men• 1,25-dihydroxyvitamin D group took it for first 2 weeks of 5-month cycle, in conjunction with a low

calcium diet• For last 6 weeks of 5-month cycle, all patients took 25-hydroxyvitamin D, 50 µg twice weekly,

together with calcium supplements to bring daily calcium intake to 1500 mg daily• 1,25-dihydroxyvitamin D group underwent treatment for mean of 2.7 years; group not receiving

1,25-dihydroxyvitamin D for mean of 3.1 years. Apparently only one or at most two patients in each group completed full 5 years. No reasons given for discontinuation of treatment

• Baseline fracture rate higher in group not receiving 1,25-dihydroxyvitamin D than in group receiving drug• Patients originally took sodium fluoride between meals.About 10% complained of nausea; this overcome

by taking it with crackers or bread• Overall, 15.9% of patients had adverse gastrointestinal or rheumatic reactions. Foot and hip pain

relieved by halving dose of sodium fluoride


213


Study Shiota, 1998194

Setting Japan




Study population Women over 50 suffering from senile or postmenopausal osteoporosis (BMD less than 0.70 g/cm2)

Recruitment procedure used Consecutive patients with diagnosis of osteoporosis visiting hospital complaining of lumbodorsal pain

Number of patients 158 or 195


Main intervention/s Ipriflavone; elcatonin; alfacalcidol + calcium; alfacalcidol + ipriflavone + calcium


Secondary outcome measures Vertebral fracture

Definition of incident Criteria of Japanese Society of Bone Mineral Researchvertebral fracture

Results: vertebral fracture Number of fractures increased in all groups but only significantly in ipriflavone and alfacalcidol + calcium groups


Quality score 6/15

Comments • Inconsistent numbers of patients given, either 158 (53, 21, 61 and 23, respectively, in four groups) or 195 (62, 25, 78 and 30, respectively)

• Not clear whether groups, as randomised, comparable: data refer only to 73 patients who completed at least 18 months but these not fully comparable in terms of age and BMD

• Only 73 women could be followed-up until at least 18 months. Reasons for withdrawal not given;compliance claimed to be relatively good in all groups except alfacalcidol + ipriflavone + calcium group

• Author argues that, as overall vertebral fracture rate only 397/1000 patient-years, compared with 760/1000 in untreated patients in another Japanese study; treatments all suppressed incidence of such fractures


Study Rozhinskaya, 199972

Setting Russia




Study population Women with established steroid-induced osteoporosis (at least two vertebral fractures + T-score < –2.5 at spine or femur neck)




Main intervention/s Monofluorophosphate + calcium, with and without alfacalcidol, compared with alfacalcidol alone

Outcome measures BMDVertebral fracturesBack pain


Results: vertebral fracture Two women (17%) in pooled fluoride groups suffered vertebral fractures, as did two (20%) in group receiving alfacalcidol alone; no statistically significant difference between fluoride and non-fluoride groups


Quality score 6/15

Comments • Patients in fluoride arms received calcium, 450 mg daily• No information given on number of withdrawals• Back pain significantly reduced in both fluoride groups but not in group taking alfacalcidol alone• All side-effects mild and transient – only one woman from fluoride groups refused to continue

treatment because of nausea and arthralgia

Appendix 4

214



Setting Japan

Date of intervention September 1995–August 1997

Source of funding Banyu Pharmaceutical Company Ltd (Tokyo, Japan)Teijin Ltd (Tokyo, Japan)

Design Double-blind, multicentre RCT

Study population Women with primary osteoporosis (lumbar T-score < –2.5, or < –1.5 with one or more vertebral fractures)

Recruitment procedure used Through 69 departments of 63 medical institutions and hospitals in Japan



Main intervention/s Alendronate compared with alfacalcidol

Primary outcome measures Effect on lumbar BMD

Secondary outcome measures Bone turnover markersAdverse effectsVertebral fracture


Results: vertebral fracture Two of 61 women in alendronate group suffered vertebral fracture as did two of 64 in alfacalcidol group.Thus no difference in fracture incidence between two groups


Quality score 11/15

Comments • Aim was to evaluate efficacy and safety of alendronate in Japanese women with osteoporosis• Alendronate compared with alfacalcidol, standard treatment for osteoporosis in Japan, as ethics

committees of many institutions would not allow use of inactive placebo• Alendronate dose half that recommended for Caucasians as, in recent dose-ranging study, this was

found to be optimal dose for Japanese population• Patients in both arms received elemental calcium, 200 mg daily• Patients allowed analgesics if in serious pain but prohibited from taking any other drugs which

affected bone or calcium metabolism during study period• Lower proportion of women had baseline fractures in alendronate (14/83, 16.9%) than in

alfacalcidol group (20/79, 25.3%), but not statistically significant• Three women (one in alendronate and two in alfacalcidol group) excluded from all analyses because

failed to take any medication.A further 44 (not attributed to groups) excluded from efficacy analyses because of protocol violations. However, vertebral fracture data available for only 125 women, 61 (58%) in alendronate and 64 (61%) in alfacalcidol group

• Similar proportion of women in each group (19/102 (18.6%) in alendronate and 25/100 (25.0%) in alfacalcidol group) reported adverse events.These occurred primarily in first 4 weeks of treatment and mostly disappeared without additional treatment after withdrawal of study drug. Gastrointestinal symptoms evenly distributed between groups

• No-one in alfacalcidol group appeared to suffer from hypercalcaemia, despite use of relatively high dose in conjunction with calcium supplementation


215



Study Thiébaud, 1994115

Setting Switzerland

Date of intervention Autumn 1988–end 1992

Source of funding Ciba-Geigy AG (donated pamidronate)

Design Small, open-label, randomised


Recruitment procedure used Consecutive postmenopausal women referred to outpatients clinic with established osteoporosis who did not like or could not take HRT



Main intervention/s Intermittent intravenous pamidronate compared with oral fluoride


Secondary outcome measures Incidence of vertebral fractureIncidence of other fracturesSide-effects

Definition of incident Greater than 25% reduction in either anterior or posterior vertebral heightvertebral fracture

Results: vertebral fracture Trend towards lower incidence in pamidronate group but numbers too small for statistical significance

Results: non-vertebral fracture Trend towards lower incidence in pamidronate group but numbers too small for statistical significance

Quality score 10/18

Comments • Authors variously describe study as ‘randomised’ and ‘partially randomised’, or that patients were ‘randomly allocated alternately’

• All women received calcium, 1 g, and vitamin D, 1000 U, daily• Compliance in pamidronate group was 100% and any side-effects minor• More side-effects in fluoride group: mainly transient arthralgias and mild gastric intolerance but

two women (12.5%) withdrew from study following stress fractures

Study Tilyard, 1992124

Setting New Zealand



Design Large, multicentre, open-label, randomised

Study population Healthy postmenopausal women with established osteoporosis (at least one non-traumatic vertebral compression fracture)

Recruitment procedure used Women diagnosed in 1986 and 1987 by 123 primary care physicians as having history of previous fracture of wrist or hip, loss of height, dowager’s hump or chronic back pain



Main intervention/s Calcitriol compared with calcium

Primary outcome measures Rate of new vertebral fractures

Secondary outcome measures Peripheral fracturesSafety of calcitriol at the study dosage

Definition of incident Reduction of 15% or more in anterior or posterior vertebral height in any 1 yearvertebral fracture

Results: vertebral fracture Significant reduction in number suffering new fractures in calcitriol versus calcium group. However, effect only evident after 2 years of treatment and only in women with mild-to-moderate osteoporosis (five or fewer vertebral fractures at baseline) or aged 65 years or older

Results: non-vertebral fracture Significant reduction in non-vertebral fractures in calcitriol group (11 fractures compared with 24 in calcium group, p < 0.05)

Quality score 16/18

Comments • Randomisation codes used to assign women to treatment groups but both patients and physicians subsequently aware of treatment assignment

• Patients given no specific instructions in relation to dietary calcium but instructed to take no calcium supplements other than those supplied for study

• Mean dietary calcium intake 880 mg/day• Of 190 (30.5%) who withdrew from study, 101 (32.2%) from calcitriol and 89 (28.9%) from calcium

group; 27 (8.6%) withdrew from calcitriol and 20 (6.5%) from calcium group due to adverse events (p > 0.05)

• Gastrointestinal symptoms led to 13 withdrawals from calcitriol and 12 from calcium group. Persistently elevated serum calcium led to two withdrawals from calcitriol but none from calcium group

• Calcium absorption status had no effect on rate of fractures in treatment groups

Appendix 4

216


Study Watts, 1990109

Setting USA




Study population Healthy postmenopausal white or Asian women with established osteoporosis (at least one but no more than four vertebral crush fractures)

Recruitment procedure used Media announcements and letters to physicians



Main intervention/s Oral etidronate, with or without phosphate

Primary outcome measures Spinal BMDIncidence of new vertebral fractures

Secondary outcome measures Incidence of non-vertebral fractures

Definition of incident Reduction of 20% or more in anterior, middle or posterior height, plus reduction of 10% or vertebral fracture more in area of previously unfractured vertebra

Results: vertebral fracture Five patients (5.1%) in etidronate-only and three (3.1%) in etidronate–phosphate group suffered new vertebral fractures, compared with seven (7.6%) in phosphate-only and ten (11.0%) in placebo group.Difference between etidronate–phosphate and placebo group statistically significant (p = 0.034). Effect greatest in subgroup with baseline BMD below 50th percentile of baseline values for total study population: of these, three (5.8%) in etidronate-only and three (6.1%) in etidronate–phosphate groups suffered new vertebral fractures, compared with seven (17.5%) in phosphate-only and ten (21.3%) in placebo groups

Results: non-vertebral fracture No apparent differences between treatment groups in number of non-vertebral fractures attributal to osteoporosis

Quality score 14/18

Comments • All patients took elemental calcium, 500 mg, for days 18–91 of 91-day cycle• All counselled in ways to achieve dietary calcium intake of at least 700 mg/day• Exceptions to stated entry criteria granted to nine women whose weight exceeded 80 kg and

to five whose age exceeded 75 years• Of 66 (15.4%) who withdrew from study, six withdrew after randomisation but before beginning of

regimen. Of remaining 60, 27 (12.8%) received etidronate and 33 (15.6%) placebo• Baseline characteristics not given for six women who withdrew immediately after randomisation• Only seven (1.6%) withdrew because of adverse events, one (0.9%) in phosphate-only, three (2.9%) in

etidronate-only, one (0.9%) in etidronate–phosphate and two (1.9%) in placebo group• Adverse effects were mild, generally infrequent and comparably distributed between treatment groups;

5–6% in all groups suffered nausea during days 1–17 (phosphate/placebo and etidronate/placebo phases of cycle); however, during days 1–3 (phosphate/placebo phase), 39% of those receiving phosphate suffered diarrhoea compared with 9% receiving placebo

• Pooling of results from etidronate-treated groups and those not receiving etidronate indicated significantly fewer etidronate-treated patients with new vertebral fractures (8 versus 17, p = 0.044)

• Pooling of results for low BMD subgroup indicated significantly fewer etidronate-treated patients with new vertebral fractures (6 versus 17, p = 0.006)

• Pooling of treatment groups and subgroup analysis in terms of BMD not pre-planned• Combination of etidronate and phosphate resulted in no apparent additional benefits beyond those

offered by etidronate alone• Following initial 2 years, patients could choose to continue original blinded treatment or take calcium

alone.a Patients completing full 3 years, whether on blinded therapy or calcium, eligible for inclusion in 2-year open-label follow-up study in which all patients took intermittent cyclical etidronate.They were then re-randomised to receive intermittent cyclical therapy with either etidronate or placebo.b

However, only results of original 2-year, double-blinded RCT included in this review.

Additional information from: a Harris, et al., 1993; b Miller, et al., 1997


217




Setting UK








Main intervention/s HRT plus etidronate, given separately and in combination




affected vertebra not considered new fracture

Results: vertebral fracture Trend to fewer vertebral fractures in treatment than control group (two in patients taking HRT alone,three in patients taking etidronate alone, one in patients taking combined therapy, and five in control group). However, numbers too small for results to be statistically significant, even when expressed as fractures/1000 patient years

Results: non-vertebral fracture No statistically significant difference between groups in terms of non-vertebral fracture

Quality score 14/18

Comments • All participants received elemental calcium, 1 g, + vitamin D2, 400 units (10 µg), daily• All participants received advice on lifestyle, dietetic modifications, and encouraged to walk about

2 miles/day• Of 14 (19.4%) who withdrew, three were from HRT, three from etidronate, four from combined therapy

and four from control group. Five withdrew as result of oestrogen-related adverse effects, two from inability to tolerate medications, five for other medical problems, one died and one was lost to follow-up.Withdrawals due to toxicity distributed as follows: HRT 3, etidronate 1, combined therapy 2,control 1

• Through all groups, 23 women complained of minor side-effects attributable to calcium but continued supplementation

• Six women (35%) taking etidronate alone complained of nausea; no-one from any other group complained of this

• Although quality of trial relatively good, numbers were too small to produce significant results in relation to fractures as opposed to BMD


219


Bisphosphonate: general information .............. 220

Bisphosphonate: methodological quality .......... 223

Vitamin D derivatives: general information ...... 225

Vitamin D derivatives: methodological quality .................................................................. 226

Calcitonin: general information ........................ 227

Calcitonin: methodological quality.................... 229

Calcium: general information ............................ 230

Calcium: methodological quality ...................... 230

Oestrogen: general information ........................ 231

Oestrogen: methodological quality.................... 231

Oestrogen-like molecules: general information ............................................ 232

Oestrogen-like molecules: methodological quality ...................................... 232

Anabolic steroids: general information ............ 233

Anabolic steroids: methodological quality ........ 233

Fluoride: general information............................ 234

Fluoride: methodological quality ...................... 236

SERMs: general information .............................. 237

SERMs: methodological quality.......................... 237

Protein supplements: general information ...... 238

Protein supplements: methodological quality .. 238

Vitamin K2: general information........................ 239

Vitamin K2: methodological quality .................. 239

Exercise: general information ............................ 239

Exercise: methodological quality ...................... 239

Comparisons with active treatments: general information ............................................ 240

Comparisons with active treatments:methodological quality ...................................... 243

Appendix 5

Summary of intervention studies*

and quality assessment

* NB. The reference numbers quoted are the principal ones for each particular trial. For details of other relevantreferences, see ‘Trials meeting the inclusion criteria’ (page 125).

Appendix 5

220 Bis

pho

spho

nate

:gen

eral

info

rmat

ion

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Ada

mi,

1995

932

BMD

(sp

ine)

Post

men

opau

sal w

omen

with

59

(48

–76)

Gro

up 1

:ora

l ale

ndro

nate

,10

mg/

day

Plac

ebo

+ e

lem

enta

l cal

cium

,500

mg

oste

opor

osis

(lu

mba

r T-s

core

> –

2),

Gro

up 2

:ora

l ale

ndro

nate

,20

mg/

day

5% h

ad v

erte

bral

frac

ture

at

entr

yA

ll pa

tient

s re

ceiv

ed e

lem

enta

l cal

cium

,500

mg/

day

Bone

,199

7942

Lum

bar

BMD

Post

men

opau

sal w

omen

with

70

.7 (

60–8

5)G

roup

1:o

ral a

lend

rona

te,1

.0 m

g/da

yPl

aceb

o +

ele

men

tal c

alci

um,5

00 m

gos

teop

aeni

a or

ost

eopo

rosi

s G

roup

2:o

ral a

lend

rona

te,2

.5 m

g/da

y(T

-sco

re <

–2

but

no m

ore

than

one

G

roup

3:o

ral a

lend

rona

te,5

.0 m

g/da

ylu

mba

r cr

ush

frac

ture

),37

% o

f who

m

All

patie

nts

rece

ived

ele

men

tal c

alci

um,5

00 m

g/da

yha

d ve

rteb

ral f

ract

ure

at e

ntry

Car

fora

,199

8952.

5BM

D*

Post

men

opau

sal w

omen

with

(4

4–73

)G

roup

1:o

ral a

lend

rona

te,5

mg/

day

Plac

ebo

+ e

lem

enta

l cal

cium

,500

mg

Vert

ebra

l fra

ctur

es*

oste

opor

osis

(lu

mba

r sp

ine

T-sc

ore

Gro

up 2

:ora

l ale

ndro

nate

,10

mg/

day

Bioc

hem

ical

mar

kers

*<

–2.

5)G

roup

3:o

ral a

lend

rona

te,2

0 m

g/da

y fo

r 15

mon

ths,

plac

ebo

for

15 m

onth

sA

ll pa

tient

s re

ceiv

ed e

lem

enta

l cal

cium

,500

mg/

day

Che

snut

,199

5962

Lum

bar

BMD

Post

men

opau

sal w

omen

with

63

(42

–75)

Gro

up 1

:ora

l ale

ndro

nate

,5 m

g/da

yPl

aceb

o +

ele

men

tal c

alci

um,5

00 m

gos

teop

aeni

a (m

axim

um lu

mba

r G

roup

2:o

ral a

lend

rona

te,1

0 m

g/da

ysp

ine

BMD

0.8

8 g/

cm2 )

but

no

Gro

up 3

:ora

l ale

ndro

nate

,20

mg/

day

for

1ye

ar,

vert

ebra

l or

hip

frac

ture

s pl

aceb

o fo

r 1

year

attr

ibut

able

to

oste

opor

osis

Gro

up 4

:ora

l ale

ndro

nate

,40

mg/

day

for

1ye

ar,

plac

ebo

for

1ye

arG

roup

5:o

ral a

lend

rona

te,4

0 m

g/da

y fo

r 3

mon

ths,

2.5

mg/

day

for

21m

onth

sA

ll pa

tient

s re

ceiv

ed e

lem

enta

l cal

cium

,500

mg/

day

Cle

mm

esen

,2

BMD

at

spin

ePo

stm

enop

ausa

l wom

en w

ith

68 (

53–8

1)G

roup

1:o

ral r

ised

rona

te,2

.5 m

g/da

yPl

aceb

o +

cal

cium

,1g/

day

1997

97es

tabl

ishe

d os

teop

oros

is (

at le

ast

Gro

up 2

:ora

l ris

edro

nate

,2.5

mg/

day

for

2 w

eeks

,on

e bu

t no

mor

e th

an fo

ur

follo

wed

by

plac

ebo

for

10 w

eeks

of a

12-

wee

k cy

cle

vert

ebra

l fra

ctur

es)

All

patie

nts

rece

ived

cal

cium

,1g/

day

FIT,

1996

982.

9 (m

ean)

Prop

ortio

n of

wom

en

Post

men

opau

sal w

omen

with

70.8

(55

–81)

Ora

l ale

ndro

nate

,5 m

g/da

y,in

crea

sed

at

Plac

ebo

Wom

en w

ith

with

new

ver

tebr

al

esta

blis

hed

oste

opor

osis

(at

leas

t24

mon

ths

to 1

0 m

g/da

ypr

e-ex

istin

g fr

actu

res

frac

ture

son

e ex

istin

g ve

rteb

ral f

ract

ure)

FIT,

1998

994.

2 (m

ean)

Prop

ortio

n of

wom

enPo

stm

enop

ausa

l wom

en w

ith67

.7 (

54–8

1)O

ral a

lend

rona

te,5

mg/

day,

incr

ease

d at

Plac

ebo

Wom

en w

ithou

t w

ith in

cide

nt n

on-

oste

opae

nia

(fem

oral

nec

k BM

D24

mon

ths

to 1

0 m

g/da

ypr

e-ex

istin

g fr

actu

res

path

olog

ical

non

-0.

68 g

/cm

2 ) bu

t no

ver

tebr

altr

aum

atic

clin

ical

fr

actu

res

frac

ture

s (n

on-

vert

ebra

l and

sy

mpt

omat

ic

vert

ebra

l fra

ctur

es)

cont

inue

d


221


Bis

pho

spho

nate

:gen

eral

info

rmat

ion

cont

d

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Har

ris,

1999

100

3In

cide

nce

of n

ew v

erte

bral

Po

stm

enop

ausa

l wom

en w

ith

69G

roup

1:o

ral r

ised

rona

te,2

.5 m

g/da

yPl

aceb

o +

ele

men

tal c

alci

um,1

g/da

yfr

actu

res

esta

blis

hed

oste

opor

osis

(ei

ther

at

Gro

up 2

:ora

l ris

edro

nate

,5 m

g/da

yIn

cide

nce

of r

adio

grap

hica

llyle

ast

two

vert

ebra

l fra

ctur

es o

r on

e A

ll pa

tient

s re

ceiv

ed e

lem

enta

l cal

cium

,1g/

day

conf

irm

ed n

on-v

erte

bral

vert

ebra

l fra

ctur

e an

d lu

mba

r-sp

ine

frac

ture

sT-

scor

e of

–2)

Cha

nges

from

bas

elin

e BM

D

Libe

rman

,199

5101

3Ef

fect

on

BMD

at

Post

men

opau

sal w

omen

with

64

(45

–80)

Gro

up 1

:ora

l ale

ndro

nate

,1.5

mg/

day

Plac

ebo

+ e

lem

enta

l cal

cium

,500

mg

lum

bar

spin

eos

teop

oros

is (

lum

bar T

-sco

re

Gro

up 2

:ora

l ale

ndro

nate

,10

mg/

day

Safe

ty a

nd t

oler

abili

ty o

f <

–2.

5) b

ut n

o ve

rteb

ral f

ract

ures

Gro

up 3

:ora

l ale

ndro

nate

,20

mg/

day,

year

s 1–

2,da

ily o

ral a

lend

rona

te5

mg/

day,

year

3Ef

fect

on

calc

ium

-reg

ulat

ing

All

patie

nts

rece

ived

ele

men

tal c

alci

um,5

00 m

g/da

yho

rmon

esEf

fect

on

bioc

hem

ical

indi

ces

of b

one

turn

over

Lind

say,

1999

102

1BM

D a

t lu

mba

r sp

ine

Post

men

opau

sal w

omen

with

61

.7 (

> 4

0)O

ral a

lend

rona

te,1

0 m

g/da

y,+

vita

min

D,4

00 IU

Plac

ebo

+ v

itam

in D

,400

IUos

teop

aeni

a al

read

y re

ceiv

ing

HRT

(T

-sco

re a

t lu

mba

r sp

ine

or fe

mor

al

neck

< –

2 an

d at

oth

er s

ite <

1.5

),57

% o

f who

m h

ad p

revi

ous

frac

ture

McC

lung

,199

8771.

5BM

D a

t lu

mba

r sp

ine

Post

men

opau

sal w

omen

with

68

Gro

up 1

:ora

l ris

edro

nate

,2.5

mg/

day

Plac

ebo

+ e

lem

enta

l cal

cium

,1g/

day

oste

opae

nia

(T-s

core

at

lum

bar

Gro

up 2

:ora

l ris

edro

nate

,5 m

g/da

ysp

ine

< –

2)A

ll pa

tient

s re

ceiv

ed e

lem

enta

l cal

cium

,1g/

day

Mon

tess

ori,

3BM

DPo

stm

enop

ausa

l wom

en w

ith

62.5

(45

–73)

Ora

l etid

rona

te,4

00 m

g/da

y,fo

r da

ys 1

–14,

+

Cal

cium

,500

mg/

day

1997

103

oste

opae

nia

(lum

bar

Z-s

core

< –

1),

calc

ium

,500

mg/

day,

for

days

15–

90 o

f 90-

day

cycl

e36

% o

f who

m h

ad v

erte

bral

frac

ture

on

ent

ry

Paci

fici,

1988

104

2Bo

ne m

iner

al c

onte

ntW

omen

with

ost

eopo

rosi

s or

58

(26

–80)

Pota

ssiu

m p

hosp

hate

,500

mg,

on d

ays

1–3,

+ o

ral

Cal

cium

car

bona

te,1

g/da

yos

teop

aeni

a (a

t le

ast

one

non-

etid

rona

te,2

00 m

g/da

y on

day

s 4–

17 o

f 73-

day

cycl

e,tr

aum

atic

ver

tebr

al fr

actu

re a

nd/o

r +

cal

cium

car

bona

te,1

g/da

yev

iden

ce o

f spi

nal d

emin

eral

isat

ion)

Pols

,199

9105

1Lu

mba

r BM

DPo

stm

enop

ausa

l wom

en w

ith

62.8

(39

–84)

Ora

l ale

ndro

nate

,10

mg/

day,

+ e

lem

enta

l cal

cium

,Pl

aceb

o +

ele

men

tal c

alci

um,5

00 m

gos

teop

aeni

a (lu

mba

r T-s

core

< –

2)50

0 m

g

Reg

inst

er,2

00010

63

Prop

ortio

n of

pat

ient

s w

ith

Post

men

opau

sal w

omen

with

71

Gro

up 1

:ora

l ris

edro

nate

,2.5

mg/

day

Plac

ebo

+ c

alci

um,1

g/da

yat

leas

t on

e in

cide

nt

esta

blis

hed

oste

opor

osis

(at

leas

t G

roup

2:o

ral r

ised

rona

te,5

mg/

day

vert

ebra

l fra

ctur

etw

o ve

rteb

ral f

ract

ures

)A

ll pa

tient

s re

ceiv

ed c

alci

um,1

g/da

y

Rei

d,19

9410

72

BMD

Post

men

opau

sal w

omen

with

66

Ora

l pam

idro

nate

,150

mg/

day,

+ e

lem

enta

l Pl

aceb

o +

ele

men

tal c

alci

um,1

g/da

yes

tabl

ishe

d os

teop

oros

is (

at le

ast

calc

ium

,1g/

day

one

vert

ebra

l fra

ctur

e)

cont

inue

d

Appendix 5

222 Bis

pho

spho

nate

:gen

eral

info

rmat

ion

cont

d

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Stor

m,1

99010

83

Bone

min

eral

con

tent

at

Post

men

opau

sal w

omen

with

68

.3 (

56–7

5)O

ral e

tidro

nate

,400

mg/

day,

for

wee

ks 1

–2 o

f Pl

aceb

o +

ele

men

tal c

alci

um,

lum

bar

spin

e an

d di

stal

es

tabl

ishe

d os

teop

oros

is (

at le

ast

15-w

eek

cycl

e +

ele

men

tal c

alci

um,5

00 m

g/da

y,50

0 m

g/da

y,+

vita

min

D,

nond

omin

ant

fore

arm

one

but

no m

ore

than

four

atr

aum

atic

+

vita

min

D,4

00 IU

/day

400

IU/d

aySp

inal

def

orm

ity in

dex

vert

ebra

l cru

sh fr

actu

res)

Loss

of h

eigh

tR

ate

of n

ew v

erte

bral

fr

actu

res

Wat

ts,1

99010

92

as R

CT

Spin

al B

MD

Post

men

opau

sal w

omen

with

65

.1G

roup

1:o

ral e

tidro

nate

,400

mg/

day,

on d

ays

4–17

,Pl

aceb

o +

ele

men

tal c

alci

um,

Rat

es o

f new

ver

tebr

al

esta

blis

hed

oste

opor

osis

(at

leas

t an

d el

emen

tal c

alci

um,5

00 m

g/da

y,on

day

s 18

-91,

500

mg/

day

on d

ays

18–9

1of

fr

actu

res

one

but

no m

ore

than

four

ver

tebr

al

of 9

1-da

y cy

cle

91-d

ay c

ycle

crus

h fr

actu

res)

Gro

up 2

:sod

ium

-pot

assi

um p

hosp

hate

,2 g

/day

on

days

1–3

,ora

l etid

rona

te,4

00 m

g/da

y on

day

s 4–

17,

and

elem

enta

l cal

cium

,500

mg/

day

on d

ays

18–9

1of

91

-day

cyc

le

Wim

alaw

ansa

,4

BMD

Post

men

opau

sal w

omen

with

64.9

(58

–72)

Gro

up 1

:ora

l etid

rona

te,4

00 m

g/da

y,fo

r w

eeks

1–2

El

emen

tal c

alci

um,1

g,+

vita

min

19

9811

0es

tabl

ishe

d os

teop

oros

is (

spin

al T

-sco

re

of 1

2-w

eek

cycl

eD

,400

U,d

aily

–2 a

nd a

t le

ast

one

but

no m

ore

than

G

roup

2:P

rem

arin

®,0

.625

mg

daily

,+ N

orge

stre

l®,

four

atr

aum

atic

tho

raci

c ve

rteb

ral

150

µg,f

or 1

2 da

ys/m

onth

+ o

ral e

tidro

nate

,cr

ush

frac

ture

s)40

0 m

g/da

y,fo

r w

eeks

1–2

of 1

2-w

eek

cycl

eA

ll pa

tient

s re

ceiv

ed e

lem

enta

l cal

cium

,1g,

+

vita

min

D,4

00 U

,dai

ly

* D

oes

not

diffe

rent

iate

bet

wee

n pr

imar

y an

d se

cond

ary

outc

ome

mea

sure

s


223


Bis

pho

spho

nate

:met

hodo

logi

cal q

ualit

y

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Ada

mi,

1995

931

13

31

09/

15 (

73)

Gro

up 1

:68

14.3

Not

spe

cifie

dG

roup

2:7

2C

ontr

ol g

roup

:71

Bone

,199

7941

13

33

314

/18

(78)

Gro

up 1

:86

36.5

Phar

mac

eutic

alG

roup

2:8

9co

mpa

nyG

roup

3:9

3C

ontr

ol g

roup

:91

Car

fora

,199

8951

11

30

17/

15 (

47)

Gro

up 1

:34

Not

spe

cifie

dN

ot s

peci

fied

Gro

up 2

:34

Gro

up 3

:34

Con

trol

gro

up:3

4

Che

snut

,199

5961

13

21

19/

18 (

50)

V1:

3218

.1Ph

arm

aceu

tical

G

roup

2:3

0co

mpa

nyG

roup

3:3

2G

roup

4:3

2G

roup

5:3

1C

ontr

ol g

roup

:31

Cle

mm

esen

,1

33

31

213

/18

(72)

Gro

up 1

:44

29.6

Not

spe

cifie

d19

9797

Gro

up 2

:44

Con

trol

gro

up:4

4

FIT,

1996

;199

8;3

33

33

318

/18

(100

)W

omen

with

pre

-N

ot s

peci

fied

Phar

mac

eutic

al

1999

exis

ting

frac

ture

:co

mpa

nyG

roup

1:1

022

Con

trol

gro

up:1

005

Wom

en w

ithou

t pr

e-ex

istin

g fr

actu

re:

Gro

up 1

:221

4C

ontr

ol g

roup

:221

8

Har

ris,

1999

100

33

33

33

18/1

8 (1

00)

Gro

up 1

:817

42Ph

arm

aceu

tical

G

roup

2:8

21(e

xclu

ding

co

mpa

nyC

ontr

ol g

roup

:820

Gro

up 1

)

Libe

rman

,199

5101

13

31

13

12/1

8 (6

7)G

roup

1:2

0216

.3Ph

arm

aceu

tical

Gro

up 2

:196

com

pany

Gro

up 3

:199

Con

trol

gro

up:3

97

Lind

say,

1999

102

11

33

11

10/1

8 (5

6)G

roup

1:2

147.

9Ph

arm

aceu

tical

Con

trol

gro

up:2

14co

mpa

ny

cont

inue

d

Appendix 5

224 Bis

pho

spho

nate

:met

hodo

logi

cal q

ualit

y co

ntd

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

McC

lung

,199

8771

13

11

07/

15 (

47)

Gro

up 1

:212

38N

ot s

peci

fied

Gro

up 2

:216

Con

trol

gro

up:2

20

Mon

tess

ori,

23

33

13

15/1

8 (8

3)G

roup

1:4

020

Phar

mac

eutic

al19

9710

3C

ontr

ol g

roup

:40

com

pany

Paci

fici,

1988

104

11

21

03

8/15

(53

)12

845

.3N

ot s

peci

fied

Pols

,199

9105

11

33

30

11/1

5 (7

3)G

roup

1:9

5011

.1Ph

arm

aceu

tical

Con

trol

gro

up:9

58co

mpa

ny

Reg

inst

er,2

00010

61

13

33

314

/18

(78)

Gro

up 1

:410

42Ph

arm

aceu

tical

Gro

up 2

:408

(exc

ludi

ngco

mpa

nyC

ontr

ol g

roup

:408

Gro

up 1

)

Rei

d,19

9410

71

12

10

38/

15 (

53)

Gro

up 1

:31

21.3

Not

spe

cifie

dC

ontr

ol g

roup

:30

Stor

m,1

99010

82

32

31

314

/18

(78)

Gro

up 1

:33

39.4

Phar

mac

eutic

alC

ontr

ol g

roup

:33

com

pany

Wat

ts,1

99010

92

33

31

314

/18

(78)

Gro

up 1

:105

15.4

Phar

mac

eutic

alG

roup

2:1

07co

mpa

nyC

ontr

ol g

roup

:104

Wim

alaw

ansa

,2

32

31

314

/18

(78)

Gro

up 1

:17

19.4

Not

spe

cifie

d19

9811

0G

roup

2:1

9C

ontr

ol g

roup

:18


225


Vit

amin

D d

eriv

ativ

es:g

ener

al in

form

atio

n

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Alo

ia,1

98811

72

BMD

*Po

stm

enop

ausa

l wom

en w

ith

64.5

Ora

l cal

citr

iol,

mea

n do

se 0

.8 µ

g/da

y,+

V

itam

in D

,400

IU/d

ayIn

cide

nce

of v

erte

bral

es

tabl

ishe

d os

teop

oros

is (

at le

ast

vita

min

D,4

00 IU

/day

frac

ture

*on

e no

n-tr

aum

atic

ver

tebr

al

Bioc

hem

ical

mea

sure

s*co

mpr

essi

on fr

actu

re)

Bone

bio

psy*

Dyk

man

,198

4631.

5Fo

rear

m b

one

mas

sR

heum

atic

dis

ease

pat

ient

s w

ith

48.7

Ora

l cal

citr

iol,

mea

n do

se 0

.25

µg/d

ay,+

Pl

aceb

o,+

ele

men

tal c

alci

um,

gluc

ocor

ticoi

d-in

duce

d os

teop

aeni

ael

emen

tal c

alci

um,5

00 m

g/da

y,+

vita

min

D,

500

mg/

day,

+ v

itam

in D

,400

IU/d

ay40

0 IU

/day

Fujit

a,19

9264

Not

spe

cifie

dVe

rteb

ral f

ract

ures

Wom

en w

ith e

stab

lishe

d os

teop

oros

is

80.3

Gro

up 1

:ora

l alfa

calc

idol

,0.7

5–1.

5 µg

/day

No

trea

tmen

t(a

t le

ast

one

non-

trau

mat

ic

Gro

up 2

:ora

l alfa

calc

idol

,0.7

5–1.

5 µg

/day

,+

vert

ebra

l fra

ctur

e)in

tram

uscu

lar

calc

itoni

n,10

uni

ts t

wic

e w

eekl

y

Gal

lagh

er,1

98911

81

(as

RC

T)

Vert

ebra

l fra

ctur

e ra

tes

Post

men

opau

sal w

omen

with

63

.2O

ral c

alci

trio

l,0.

5 µg

/day

,inc

reas

ed t

o 0.

75 o

r Pl

aceb

oes

tabl

ishe

d os

teop

oros

is (

at le

ast

1.0

µg/d

ay a

t in

vest

igat

or’s

disc

retio

non

e no

n-tr

aum

atic

ver

tebr

al fr

actu

re)

Gal

lagh

er,1

990a

119

2Sa

fety

Post

men

opau

sal w

omen

with

69

.7O

ral c

alci

trio

l,m

ean

dose

0.6

2 µg

/day

,+

Plac

ebo

+ v

itam

in D

2,40

0 IU

/day

BMD

esta

blis

hed

oste

opor

osis

(at

leas

t vi

tam

in D

2,40

0 IU

/day

one

non-

trau

mat

ic v

erte

bral

frac

ture

)

Ori

mo,

1987

120

1.7–

2.1

(mea

n)Ve

rteb

ral c

rush

frac

ture

sW

omen

with

est

ablis

hed

seni

le

71.7

Gro

up 1

:ora

l alfa

calc

idol

,1µg

/day

Con

trol

gro

up 1

:no

trea

tmen

tos

teop

oros

is (

decr

ease

d ve

rteb

ral

Gro

up 2

:ora

l alfa

calc

idol

,1µg

/day

+ c

alci

um,

Con

trol

gro

up 2

:cal

cium

,1g/

day

dens

ity a

nd a

t le

ast

one

crus

h fr

actu

re)

1g/

day

Ori

mo,

1994

121

1BM

D*

Post

men

opau

sal w

omen

with

71.9

Ora

l alfa

calc

idol

,1µg

/day

,+ e

lem

enta

l cal

cium

,Pl

aceb

o +

ele

men

tal c

alci

um,

New

ver

tebr

al fr

actu

res*

esta

blis

hed

oste

opor

osis

(62

.5%

had

300

mg/

day

300

mg/

day

Non

-ver

tebr

al fr

actu

res*

frac

ture

of s

pine

,fem

ur n

eck

or

radi

us a

t en

try)

Ott

,198

9122

2C

hang

e in

bon

e m

ass

Post

men

opau

sal w

omen

with

67

.5O

ral c

alci

trio

l,m

ean

dose

0.4

3 µg

/day

Plac

ebo

esta

blis

hed

oste

opor

osis

(at

leas

t tw

o no

n-tr

aum

atic

ver

tebr

al

com

pres

sion

frac

ture

s)

Shir

aki,

1996

682

BMD

Post

men

opau

sal w

omen

with

dia

gnos

is

72.4

Ora

l alfa

calc

idol

,0.7

5 µg

/day

+ e

lem

enta

l Pl

aceb

o +

ele

men

tal c

alci

um,

Vert

ebra

l and

non

-of

pro

babl

e or

def

inite

ost

eopo

rosi

s ca

lciu

m,3

00 m

g/da

y30

0 m

g/da

yve

rteb

ral f

ract

ures

(49%

had

at

leas

t on

e ve

rteb

ral

frac

ture

at

entr

y)

* D

oes

not

diffe

rent

iate

bet

wee

n pr

imar

y an

d se

cond

ary

outc

ome

mea

sure

s

Appendix 5

226 Vit

amin

D d

eriv

ativ

es:m

etho

dolo

gica

l qua

lity

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Alo

ia,1

9881

171

32

10

18/

15 (

53)

Gro

up 1

:17

20.6

Not

spe

cifie

dC

ontr

ol g

roup

:17

Dyk

man

,198

4631

12

13

19/

18 (

50)

3023

.3N

ot s

peci

fied

Fujit

a,19

9264

13

22

03

11/1

5 (7

3)G

roup

1:8

6.2

Not

spe

cifie

dG

roup

2:8

Con

trol

gro

up:8

Gal

lagh

er,1

98911

81

32

10

29/

15 (

60)

Gro

up 1

:38

12.7

Res

earc

h bo

dy/

Con

trol

gro

up:3

3ph

arm

aceu

tical

com

pany

Gal

lagh

er,1

990a

119

31

21

02

9/15

(60

)G

roup

1:2

520

Phar

mac

eutic

al c

ompa

nyC

ontr

ol g

roup

:25

Ori

mo,

1987

120

13

12

03

10/1

5 (6

7)G

roup

1:2

2N

ot s

peci

fied

Not

spe

cifie

dG

roup

2:1

6C

ontr

ol g

roup

1:2

3C

ontr

ol g

roup

2:2

5

Ori

mo,

1994

121

23

21

13

12/1

8 (6

7)G

roup

1:3

87.

5N

ot s

peci

fied

Con

trol

gro

up:4

2

Ott

,198

9122

31

23

12

12/1

8 (6

7)G

roup

1:4

316

.3R

esea

rch

body

/C

ontr

ol g

roup

:43

phar

mac

eutic

al c

ompa

ny

Shir

aki,

1996

682

31

31

313

/18

(72)

Gro

up 1

:57

30.1

(?)

Not

spe

cifie

dC

ontr

ol g

roup

:56


227


Cal

cito

nin:

gene

ral i

nfo

rmat

ion

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Ada

mi,

1995

932

BMD

(sp

ine)

Post

men

opau

sal w

omen

with

59

(48

–76)

Intr

anas

al c

alci

toni

n,10

0 IU

/day

,+ o

ral

Ora

l ele

men

tal c

alci

um,5

00 m

g/da

yos

teop

oros

is (

lum

bar T

-sco

re >

–2)

;el

emen

tal c

alci

um,5

00 m

g/da

y5%

had

ver

tebr

al fr

actu

re a

t en

try

Agr

awal

,198

1612

Tota

l bod

y ca

lciu

m*

Men

with

est

ablis

hed

oste

opor

osis

Su

bcut

aneo

us c

alci

toni

n,10

0 M

RC

uni

ts/d

ay,+

C

ontr

ol g

roup

1:o

ral c

alci

um,

Bone

min

eral

con

tent

*(a

t le

ast

one

non-

trau

mat

ic v

erte

bral

or

al c

alci

um,1

g/da

y,+

vita

min

D2,

500

IU/d

ay1

g/da

y,+

vita

min

D2,

500

IU/d

ayN

ew v

erte

bral

eve

nts*

com

pres

sion

frac

ture

)C

ontr

ol g

roup

2:v

itam

in D

2,Bi

oche

mic

al in

dice

s*80

0 IU

/day

Peri

pher

al fr

actu

res*

Cri

stal

lini,

1993

128

2BM

DPo

stm

enop

ausa

l wom

en w

ith

63.5

(41

–73)

Gro

up 1

:cyc

lical

intr

amus

cula

r ca

lcito

nin,

Cho

leca

lcife

rol,

0.25

µg/

day,

Vert

ebra

l fra

ctur

eses

tabl

ishe

d os

teop

oros

is (

at le

ast

100

U/d

ay,+

pot

assi

um p

hosp

hate

,2 g

/day

,+

ele

men

tal c

alci

um,1

g/da

ytw

o no

n-tr

aum

atic

ver

tebr

al fr

actu

res)

days

1–3

0;or

al c

hole

calc

ifero

l,0.

25 µ

g/da

y,+

el

emen

tal c

alci

um,1

g/da

y,da

ys 3

1–75

of

90-d

ay c

ycle

Gro

up 2

:cyc

lical

intr

amus

cula

r ca

lcito

nin,

100

U/d

ay,+

pot

assi

um p

hosp

hate

,2 g

/day

,da

ys 1

–30;

chol

ecal

cife

rol,

0.25

µg/

day,

+

elem

enta

l cal

cium

,1g/

day,

+ s

odiu

m fl

uori

de,

40 m

g/da

y,da

ys 3

1–75

of 9

0-da

y cy

cle

Elle

ring

ton,

2BM

D o

f lum

bar

spin

ePo

stm

enop

ausa

l wom

en w

ith

55.8

(48

–64)

Gro

up 1

:int

rana

sal c

alci

toni

n,10

0 IU

/day

Plac

ebo

1996

129

and

hip

oste

opae

nia

(T-s

core

in s

pine

or

Gro

up 2

:int

rana

sal c

alci

toni

n,20

0 IU

hi

p <

–1.

2)3

times

wee

kly

Fujit

a,19

9264

Not

spe

cifie

dVe

rteb

ral f

ract

ures

Wom

en w

ith e

stab

lishe

d os

teop

oros

is

79G

roup

1:i

ntra

mus

cula

r ca

lcito

nin,

10 u

nits

N

o tr

eatm

ent

(at

leas

t on

e no

n-tr

aum

atic

tw

ice

wee

kly

vert

ebra

l fra

ctur

e)G

roup

2:i

ntra

mus

cula

r ca

lcito

nin,

10 u

nits

tw

ice

wee

kly,

+ a

lfaca

lcid

ol,0

.75–

1.5

µg/d

ay

Gen

nari

,198

5651

Bone

min

eral

con

tent

Po

stm

enop

ausa

l wom

en w

ith

58.7

(44

–70)

Gro

up 1

:sub

cuta

neou

s ca

lcito

nin,

100

MR

C u

nits

,C

alci

um,1

g/da

yof

lum

bar

spin

e an

d es

tabl

ishe

d os

teop

oros

is (

at le

ast

on a

ltern

ate

days

fem

oral

dia

phys

is*

two

non-

trau

mat

ic v

erte

bral

G

roup

2:s

ubcu

tane

ous

calc

itoni

n,10

0 M

RC

Ve

rteb

ral f

ract

ures

*co

mpr

essi

on fr

actu

res)

units

/day

Seru

m a

nd u

rina

ry

All

patie

nts

rece

ived

ora

l cal

cium

,1g/

day

para

met

ers*

Hiz

met

li,19

9813

02

Lum

bar

and

fem

oral

Po

stm

enop

ausa

l wom

en n

ewly

58

.0G

roup

1:i

ntra

nasa

l cal

cito

nin,

50 IU

/day

Elem

enta

l cal

cium

,100

0 m

g/da

y,+

ne

ck B

MD

diag

nose

d as

ost

eopo

rotic

G

roup

2:i

ntra

nasa

l cal

cito

nin,

100

IU/d

ayvi

tam

in D

,400

IU/d

ay o

r Ve

rteb

ral f

ract

ure

(T-s

core

< –

2.5)

All

patie

nts

rece

ived

ele

men

tal c

alci

um,

50,0

00 IU

/wee

k10

00 m

g/da

y,+

vita

min

D,4

00 IU

/day

or

50,0

00 IU

/wee

k

cont

inue

d

Appendix 5

228 Cal

cito

nin:

gene

ral i

nfo

rmat

ion

cont

d

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Hod

sman

,199

7131

2C

hang

e in

BM

D a

t Po

stm

enop

ausa

l wom

en w

ith

67H

uman

par

athy

roid

hor

mon

e,80

0 IU

equ

ival

ent/

hPT

H,8

00 IU

equ

ival

ents

/day

,,lu

mba

r sp

ine

esta

blis

hed

oste

opor

osis

(at

leas

t da

y,da

ys 1

–28;

subc

utan

eous

cal

cito

nin,

75 IU

/day

,da

ys 1

–28;

plac

ebo

days

29–

70;

one

vert

ebra

l com

pres

sion

frac

ture

)da

ys 2

9–70

;ele

men

tal c

alci

um;5

00 m

g/da

y,el

emen

tal c

alci

um,5

00 m

g/da

y,da

ys 7

1–90

of 9

0-da

y cy

cle

days

71–

90 o

f 90-

day

cycl

e

Ove

rgaa

rd,1

99260

2Bo

ne m

iner

al c

onte

nt

Elde

rly

wom

en w

ith m

oder

ate

70 (

68–7

2)G

roup

1:i

ntra

nasa

l cal

cito

nin,

50 IU

/day

Plac

ebo

+ c

alci

um,5

00 m

g/da

yof

dis

tal f

orea

rm a

nd

oste

opor

osis

(T-

scor

e at

dis

tal

Gro

up 2

:int

rana

sal c

alci

toni

n,10

0 IU

/day

lum

bar

spin

efo

rear

m <

–2)

;6%

had

ver

tebr

al

Gro

up 3

:int

rana

sal c

alci

toni

n,20

0 IU

/day

Rat

es o

f ver

tebr

al a

nd

frac

ture

s at

ent

ryA

ll pa

tient

s re

ceiv

ed o

ral c

alci

um,5

00 m

g/da

ype

riph

eral

frac

ture

s

Pont

irol

i,19

9113

20.

5Pa

inW

omen

with

pos

tmen

opau

sal o

r 71

(60–

83)

Intr

anas

al c

alci

toni

n,10

0 IU

,on

alte

rnat

e da

ysIn

tram

uscu

lar

calc

itoni

n,10

0 IU

,on

Met

abol

ic in

dice

sse

nile

ost

eopo

rosi

s (a

t le

ast

1al

tern

ate

days

Bone

min

eral

con

tent

vert

ebra

l fra

ctur

e)

PRO

OF

stud

y785

Inci

denc

e of

new

Po

stm

enop

ausa

l wom

en w

ith

68.3

(45

–95)

Gro

up 1

:int

rana

sal c

alci

toni

n,10

0 IU

/day

Plac

ebo

calc

itoni

n +

cal

cium

,ve

rteb

ral f

ract

ures

esta

blis

hed

oste

opor

osis

(at

leas

t on

e G

roup

2:i

ntra

nasa

l cal

cito

nin,

200

IU/d

ay10

00 m

g/da

y,+

vita

min

D,4

00 IU

/day

but

not

mor

e th

an fi

ve n

on-t

raum

atic

G

roup

3:c

alci

toni

n,40

0 IU

/day

vert

ebra

l fra

ctur

es,a

nd T

-sco

re a

t A

ll pa

tient

s re

ceiv

ed c

alci

um,1

000

mg/

day,

+

lum

bar

spin

e <

–2)

vita

min

D,4

00 IU

/day

Ric

o,19

9213

42

Vert

ebra

l fra

ctur

e ra

tePo

stm

enop

ausa

l wom

en w

ith

68.3

Intr

amus

cula

r ca

lcito

nin,

100

IU/d

ay,f

or

Elem

enta

l cal

cium

,500

mg/

day,

for

esta

blis

hed

oste

opor

osis

(at

leas

t on

e 10

day

s/m

onth

+ o

ral e

lem

enta

l cal

cium

,10

day

s/m

onth

non-

trau

mat

ic v

erte

bral

cru

sh fr

actu

re)

500

mg/

day,

for

10 d

ays/

mon

th

Ric

o,19

9513

52

Spin

al d

efor

mity

Post

men

opau

sal w

omen

with

69

.2In

tram

uscu

lar

calc

itoni

n,10

0 IU

/day

,for

El

emen

tal c

alci

um,5

00 m

g/da

y,fo

r Ve

rteb

ral f

ract

ure

esta

blis

hed

oste

opor

osis

(m

ore

than

10

day

s/m

onth

+ o

ral e

lem

enta

l cal

cium

,10

day

s/m

onth

one

non-

trau

mat

ic v

erte

bral

frac

ture

)50

0 m

g/da

y,fo

r 10

day

s/m

onth

Rin

ge,1

98713

60.

5Ve

rteb

ral f

ract

ures

/ M

en a

nd w

omen

with

‘inc

ipie

nt t

o 49

.5Su

bcut

aneo

us c

alci

toni

n,10

0 IU

,on

alte

rnat

e da

ysN

o tr

eatm

ent

defo

rmiti

es*

seve

re s

igns

’ of s

tero

id-in

duce

d Pa

in*

oste

opor

osis

Adv

erse

effe

cts*

Non

-ver

tebr

al fr

actu

re*

Rin

ge,1

99013

71

Pain

*M

en a

nd w

omen

with

pri

mar

y 60

.8G

roup

1:s

ubcu

tane

ous

calc

itoni

n,10

0 IU

/day

Cal

cium

,1g/

day

Vert

ebra

l and

non

-os

teop

oros

is (

unde

fined

)G

roup

2:s

ubcu

tane

ous

calc

itoni

n,10

0 IU

,ve

rteb

ral f

ract

ures

*on

alte

rnat

e da

ysBM

D*

All

patie

nts

rece

ived

cal

cium

,1g/

day

Bone

bio

psy*

* D

oes

not

diffe

rent

iate

bet

wee

n pr

imar

y an

d se

cond

ary

outc

ome

mea

sure

s


229


Cal

cito

nin:

met

hodo

logi

cal q

ualit

y

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Ada

mi,

1995

931

13

31

09/

15 (

73)

Gro

up 1

:68

14.3

Not

spe

cifie

dR

Gro

up 2

:72

Con

trol

gro

up:7

1

Agr

awal

,198

1611

32

10

18/

15 (

53)

3943

.6R

esea

rch

body

/ ph

arm

aceu

tical

com

pany

Cri

stal

lini,

1993

128

11

21

01

6/15

(40

)G

roup

1:4

038

.4N

ot s

peci

fied

Gro

up 2

:40

Con

trol

gro

up:4

5

Elle

ring

ton,

1996

129

11

23

03

10/1

5 (6

7)G

roup

1:3

617

.1R

esea

rch

body

/G

roup

2:3

5ph

arm

aceu

tical

com

pany

Con

trol

gro

up:4

6

Fujit

a,19

9264

13

22

03

11/1

5 (7

3)G

roup

1:8

6.2

Not

spe

cifie

dG

roup

2:8

Con

trol

gro

up:8

Gen

nari

,198

5651

12

10

16/

15 (

40)

8267

%N

ot s

peci

fied

Hiz

met

li,19

9813

01

12

30

29/

15 (

73)

Gro

up 1

:35

18.7

Not

spe

cifie

dG

roup

2:4

1C

ontr

ol g

roup

:31

Hod

sman

,199

7131

11

23

13

11/1

8 (6

1)39

23.1

Res

earc

h bo

dy/

phar

mac

eutic

al c

ompa

ny

Ove

rgaa

rd,1

99260

13

23

33

15/1

8 (8

3)G

roup

1:5

215

.4R

esea

rch

body

/G

roup

2:5

2ph

arm

aceu

tical

com

pany

Gro

up 3

:52

Con

trol

gro

up:5

2

Pont

irol

i,19

9113

11

12

10

16/

15 (

40)

Gro

up 1

:720

Not

spe

cifie

dC

ontr

ol g

roup

:8

PRO

OF

stud

y781

33

10

311

/15

(73)

Gro

up 1

:316

9.8

Not

spe

cifie

dG

roup

2:3

16G

roup

3:3

12C

ontr

ol g

roup

:311

Ric

o,19

9213

41

32

31

313

/18

(72)

Gro

up 1

:32

5G

over

nmen

t bo

dyC

ontr

ol g

roup

:28

cont

inue

d

Appendix 5

230 Cal

cito

nin:

met

hodo

logi

cal q

ualit

y co

ntd

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Ric

o,19

9513

51

32

31

313

/18

(72)

Gro

up 1

:36

5.6

Not

spe

cifie

dC

ontr

ol g

roup

:36

Rin

ge,1

98713

61

12

31

19/

18 (

50)

Gro

up 1

:19

5.3

Not

spe

cifie

dC

ontr

ol g

roup

:19

Rin

ge,1

99013

71

12

11

17/

18 (

39)

6711

.9N

ot s

peci

fied

Cal

cium

:met

hodo

logi

cal q

ualit

y

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Gut

teri

dge,

1993

801

11

10

15/

15 (

33)

Gro

up 1

:8N

ot s

peci

fied

Not

spe

cifie

dC

ontr

ol g

roup

:9

Rec

ker,

1996

142

13

33

02

12/1

5 (8

0)Pr

eval

ent

frac

ture

s gr

oup

Not

spe

cifie

dD

airy

indu

stry

/ G

roup

1:5

3go

vern

men

t bo

dy/

Con

trol

gro

up:4

1ph

arm

aceu

tical

com

pany

Cal

cium

stu

dies

:gen

eral

info

rmat

ion

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Gut

teri

dge,

1993

80N

ot c

lear

,BM

D*

Post

men

opau

sal w

omen

with

N

ot s

peci

fied

Cal

cium

,1g/

day,

+ v

itam

in D

2,0.

5 m

g N

o tr

eatm

ent

but

over

1Ve

rteb

ral f

ract

ures

*es

tabl

ishe

d os

teop

oros

is (

at le

ast

(20,

000

IU)/

wee

kon

e ve

rteb

ral f

ract

ure

follo

win

g m

inor

tra

uma)

Rec

ker,

1996

142

4.3

(mea

n)In

cide

nce

of v

erte

bral

El

derl

y w

omen

with

low

sel

f-cho

sen

74.9

Cal

cium

,1.2

g/d

ayPl

aceb

ofr

actu

res

calc

ium

inta

kes

and

prev

alen

t Fo

rear

m b

one

mas

s ch

ange

sve

rteb

ral f

ract

ures

* D

oes

not

diffe

rent

iate

bet

wee

n pr

imar

y an

d se

cond

ary

outc

ome

mea

sure

s


231


Oes

trog

en:g

ener

al in

form

atio

n

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Lufk

in,1

99217

81

Bone

tur

nove

r as

sess

ed b

y Po

stm

enop

ausa

l wom

en w

ith

64.8

(47

–75)

Tran

sder

mal

17β

oest

radi

ol,0

.1m

g,on

day

s 1–

21Pl

aceb

obi

oche

mic

al m

arke

rs a

nd

esta

blis

hed

oste

opor

osis

(lo

w B

MD

+

ora

l med

roxy

prog

este

rone

ace

tate

,10

mg,

on

iliac

bon

e hi

sto-

and

at le

ast

one

vert

ebra

l fra

ctur

e)da

ys 1

1–21

of 2

8-da

y cy

cle

mor

phom

etry

*

BMD

*

Vert

ebra

l fra

ctur

e ra

te*

Paci

fici,

1988

104

2Bo

ne m

iner

al c

onte

ntW

omen

with

ost

eopo

rosi

s or

58 (

26–8

0)C

onju

gate

d oe

stro

gens

,0.6

25 m

g/da

y,fo

r C

alci

um c

arbo

nate

,100

0 m

g/da

yos

teop

aeni

a (a

t le

ast

one

non-

trau

mat

ic

25 d

ays/

mon

th,m

edro

xypr

oges

tero

ne,1

0 m

g/da

y,ve

rteb

ral f

ract

ure

and/

or e

vide

nce

on d

ays

15–2

5 +

cal

cium

car

bona

te,1

000

mg/

day

of s

pina

l dem

iner

alis

atio

n)

Wim

alaw

ansa

,4

BMD

Post

men

opau

sal w

omen

with

64.9

(58

–72)

Prem

arin

®,0

.625

mg/

day,

daily

+ N

orge

stre

l®,

Elem

enta

l cal

cium

,1g/

day,

+

1998

110

esta

blis

hed

oste

opor

osis

(sp

inal

150

µg/d

ay,f

or 1

2 da

ys/m

onth

+ e

lem

enta

l vi

tam

in D

2,40

0 U

/day

T-sc

ore

–2 a

nd a

t le

ast

one

but

no

calc

ium

,1g/

day

+ v

itam

in D

2,40

0 U

/day

mor

e th

an fo

ur a

trau

mat

ic t

hora

cic

vert

ebra

l cru

sh fr

actu

res)

Zar

cone

,199

7144

64 m

onth

sBM

D a

t lu

mba

r sp

ine

Post

men

opau

sal w

omen

with

(4

5–63

)G

roup

1:c

onju

gate

d eq

uine

oes

trog

en,0

.15

mg/

day

Plac

ebo

Vert

ebra

l fra

ctur

esos

teop

oros

is (

max

imum

BM

D

Gro

up 2

:con

juga

ted

equi

ne o

estr

ogen

,0.3

mg/

day

at lu

mba

r sp

ine

0.88

g/c

m2 )

Gro

up 3

:con

juga

ted

equi

ne o

estr

ogen

,0.6

25 m

g/da

y

* D

oes

not

diffe

rent

iate

bet

wee

n pr

imar

y an

d se

cond

ary

outc

ome

mea

sure

s

Oes

trog

en:m

etho

dolo

gica

l qua

lity

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Lufk

in,1

99217

81

13

30

311

/15

(73)

Gro

up 1

:36

10.7

Phar

mac

eutic

al c

ompa

nyC

ontr

ol g

roup

:39

Paci

fici,

1988

104

11

21

03

8/15

(53

)12

845

.3N

ot s

tate

d

Wim

alaw

ansa

,2

32

31

314

/18

(78)

Gro

up 1

:17

19.4

Not

sta

ted

1998

110

Gro

up 2

:19

Con

trol

gro

up:1

8

Zar

cone

,199

7144

11

21

11

7/18

(39

)13

29

Not

sta

ted

Appendix 5

232 Oes

trog

en-l

ike

mo

lecu

les:

gene

ral i

nfo

rmat

ion

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Mau

geri

,199

4157

2BM

DEl

derl

y w

omen

with

est

ablis

hed

69.9

Ipri

flavo

ne,6

00 m

g/da

y,+

cal

cium

,1g/

day

Plac

ebo

+ c

alci

um,1

g/da

yBo

ne m

etab

olis

m

oste

opor

osis

(at

leas

t on

e ve

rteb

ral

mar

ker

para

met

ers

frac

ture

,and

T-s

core

at

dist

al r

adiu

s of

< –

2)

Pass

eri,

1992

158

1Bo

ne m

ass

Elde

rly

wom

en w

ith e

stab

lishe

d69

(65

–85)

Ipri

flavo

ne,2

00 m

g/da

y,+

cal

cium

(ca

rbon

ate

Plac

ebo

+ c

alci

um (

carb

onat

e an

d Bi

omar

kers

oste

opor

osis

(at

leas

t on

e ve

rteb

ral

and

gluc

onat

e),1

g/da

ygl

ucon

ate)

,1g/

day

frac

ture

,and

T-s

core

at

dist

al r

adiu

s of

< –

2)

Pass

eri,

1995

159

2BM

DEl

derl

y w

omen

with

est

ablis

hed

69.5

(65

–79)

Ipri

flavo

ne,6

00 m

g/da

y,+

ele

men

tal c

alci

um,

Plac

ebo

+ e

lem

enta

l cal

cium

,1g/

day

Safe

tyos

teop

oros

is (

at le

ast

one

vert

ebra

l 1

g/da

yfr

actu

re,a

nd fo

rear

m Z

-sco

re o

f < –

2)

Oes

trog

en-l

ike

mo

lecu

les:

met

hodo

logi

cal q

ualit

y

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Mau

geri

,199

4157

11

22

01

7/15

(47

)10

016

Not

spe

cifie

d

Pass

eri,

1992

158

13

33

01

11/1

5 (7

3)G

roup

1:1

44

Not

spe

cifie

dC

ontr

ol g

roup

:14

Pass

eri,

1995

159

13

33

32

15/1

8 (8

3)G

roup

1:2

544

.9Ph

arm

aceu

tical

com

pany

Con

trol

gro

up:2

4


233


Ana

bolic

ste

roid

s:ge

nera

l inf

orm

atio

n

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Che

snut

,198

39629

mon

ths

Tota

l bod

y ca

lciu

mPo

stm

enop

ausa

l wom

en w

ith v

erte

bral

68.7

(50

–80)

Stan

ozol

ol,6

mg,

3 w

eeks

in 4

Plac

ebo

Bone

mas

sos

teop

aeni

a or

one

or

mor

eBi

oche

mic

al m

arke

rsat

raum

atic

spi

nal c

ompr

essi

on fr

actu

res

Ana

bolic

ste

roid

s:m

etho

dolo

gica

l qua

lity

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Che

snut

,198

3961

13

30

210

/15

(67)

Gro

up 1

:23

18.6

Phar

mac

eutic

al c

ompa

ny/

Con

trol

gro

up:2

3go

vern

men

t bo

dy

Appendix 5

234 Flu

ori

de:g

ener

al in

form

atio

n

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Broc

kste

dt,1

99673

3BM

D*

Wom

en w

ith o

steo

poro

sis

67.2

Fluo

ride

(as

sod

ium

mon

oflu

orop

hosp

hate

),Fl

uori

de (

as s

odiu

m

Bone

mar

kers

*26

.4 m

g/da

y,fo

r 3

mon

ths

in e

very

6 +

cal

cium

m

onof

luor

opho

spha

te),

26.4

mg/

day,

+

Vert

ebra

l fra

ctur

es*

carb

onat

e,10

00 m

g/da

yca

lciu

m c

arbo

nate

,100

0 m

g/da

yA

dver

se e

vent

s*

Gut

teri

dge,

1993

80N

ot c

lear

,BM

D*

Post

men

opau

sal w

omen

with

N

ot s

peci

fied

Ente

ric-

coat

ed s

odiu

m fl

uori

de,6

0 m

g/da

y,fo

r C

ontr

ol g

roup

1:c

alci

um,1

g/da

y,+

bu

t ov

er 1

Vert

ebra

l fra

ctur

es*

esta

blis

hed

oste

opor

osis

(at

leas

t m

onth

s 1–

6 of

9-m

onth

cyc

le +

cal

cium

,1g/

day,

+

vita

min

D2,

0.5

mg

(20,

000

IU)/

wee

kon

e ve

rteb

ral f

ract

ure

follo

win

g vi

tam

in D

2,0.

5 m

g (2

0,00

0 IU

)/w

eek

Con

trol

gro

up 2

:no

trea

tmen

tm

inor

tra

uma)

Gut

teri

dge,

1996

792.

25BM

D a

t lu

mba

r sp

ine,

Post

men

opau

sal w

omen

with

N

ot s

peci

fied

Ente

ric-

coat

ed s

odiu

m fl

uori

de,6

0 m

g/da

y,fo

r C

alci

um,1

g/da

y,+

vita

min

D2,

uppe

r fe

mur

and

low

er

esta

blis

hed

oste

opor

osis

(at

leas

t m

onth

s 1–

6 of

9-m

onth

cyc

le +

cal

cium

,1g/

day,

+

0.25

–0.5

mg/

wee

ktib

ia a

nd fi

bula

*on

e ve

rteb

ral f

ract

ure)

vita

min

D2,

0.25

–0.5

mg/

wee

kIn

cide

nt v

erte

bral

frac

ture

s*

Han

sson

,198

7167

3Sp

inal

bon

e m

iner

al c

onte

ntPo

stm

enop

ausa

l wom

en w

ith

65.9

Gro

up 1

:sod

ium

fluo

ride

,30

mg/

day

Con

trol

gro

up 1

:cal

cium

es

tabl

ishe

d os

teop

oros

is (

at le

ast

Gro

up 2

:sod

ium

fluo

ride

,10

mg/

day

(com

bina

tion

of b

icar

bona

te,l

acta

te

one

and

max

imum

of t

hree

ver

tebr

al

All

patie

nts

rece

ived

cal

cium

,1g/

day

and

gluc

onat

e),1

g/da

yco

mpr

essi

on fr

actu

res

sust

aine

d du

ring

(c

ombi

natio

n of

bic

arbo

nate

,lac

tate

C

ontr

ol g

roup

2:p

lace

bom

inor

tra

uma)

and

gluc

onat

e)

Kle

erek

oper

,4

Vert

ebra

l fra

ctur

ePo

stm

enop

ausa

l wom

en w

ith

70.0

Sodi

um fl

uori

de,7

5 m

g/da

y (a

vera

ge in

itial

dos

e),

Plac

ebo

+ c

alci

um,1

500

mg/

day

1991

168

esta

blis

hed

oste

opor

osis

(at

leas

t +

cal

cium

,150

0 m

g/da

yon

e ve

rteb

ral c

ompr

essi

on fr

actu

re

or t

wo

or m

ore

non-

cont

iguo

us

vert

ebra

l wed

ge d

efor

miti

es)

Meu

nier

,199

8169

2Pe

rcen

tage

of p

atie

nts

with

Po

stm

enop

ausa

l wom

en w

ith

65.7

(47

–76)

Gro

up 1

:sod

ium

fluo

ride

,50

mg/

day

Cal

cium

car

bona

te,1

g/da

y,at

leas

t on

e ne

w v

erte

bral

es

tabl

ishe

d os

teop

oros

is (

at le

ast

Gro

up 2

:mon

oflu

orop

hosp

hate

,150

mg/

day

+ v

itam

in D

2,80

0 IU

/day

frac

ture

bet

wee

n T

4 an

d on

e ve

rteb

ral f

ract

ure)

Gro

up 3

:mon

oflu

orop

hosp

hate

,200

mg/

day

L5 a

fter

2 y

ears

All

patie

nts

rece

ived

cal

cium

car

bona

te,1

g/da

y,+

vita

min

D2,

800

IU/d

ay

Pak,

1995

170

4In

cide

nt v

erte

bral

frac

ture

sPo

stm

enop

ausa

l wom

en w

ith

67.6

(55

–81)

Slow

-rel

ease

sod

ium

fluo

ride

,50

mg/

day,

Plac

ebo

+ c

alci

um c

itrat

e,80

0 m

g/da

yes

tabl

ishe

d os

teop

oros

is (

at le

ast

+ c

alci

um c

itrat

e,80

0 m

g/da

yon

e no

n-tr

aum

atic

ver

tebr

al fr

actu

re)

Reg

inst

er,1

99817

14

Num

ber

of p

atie

nts

with

Po

stm

enop

ausa

l wom

en w

ith

63So

dium

mon

oflu

orop

hosp

hate

,152

mg/

day,

+

Cal

cium

car

bona

te,2

500

mg/

day

new

ver

tebr

al fr

actu

res

mod

erat

e os

teop

oros

is (

lum

bar

calc

ium

car

bona

te,2

500

mg/

day

(equ

ival

ent

to

(equ

ival

ent

to 1

000

mg/

day

calc

ium

)du

ring

the

4-y

ear

spin

e T-

scor

e of

–2.

5),2

% o

f who

m

calc

ium

,100

0 m

g/da

y)tr

eatm

ent

peri

odha

d ve

rteb

ral f

ract

ures

at

entr

y

Rig

gs,1

99017

24

Vert

ebra

l fra

ctur

esPo

stm

enop

ausa

l wom

en w

ith68

(m

edia

n)

Sodi

um fl

uori

de,7

5 m

g/da

y,+

ele

men

tal c

alci

um,

Plac

ebo

+ e

lem

enta

l cal

cium

,Si

de-e

ffect

s,in

clud

ing

esta

blis

hed

oste

opor

osis

(lo

w B

MD

(50–

75)

1500

mg/

day

1500

mg/

day

non-

vert

ebra

l fra

ctur

esat

lum

bar

spin

e an

d at

leas

t on

e ve

rteb

ral f

ract

ure)

cont

inue

d


235


Flu

ori

de:g

ener

al in

form

atio

n co

ntd

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Rin

ge,1

99817

33

BMD

of l

umba

r sp

ine

Men

with

ear

ly id

iopa

thic

ost

eo-

53 (

33–6

8)So

dium

mon

oflu

orop

hosp

hate

,114

mg/

day,

+

Cal

cium

,100

0 m

g/da

ypo

rosi

s (T

-sco

re <

–2.

5 bu

t no

ca

lciu

m,5

00 m

g/da

y fo

r 3

mon

ths

in 4

,si

gnifi

cant

def

orm

ity o

r si

gns

of

1000

mg/

day

for

rem

aini

ng m

onth

prev

ious

ver

tebr

al fr

actu

res)

Rin

ge,1

99917

43

BMD

Post

men

opau

sal w

omen

with

64.0

Gro

up 1

:mon

oflu

orop

hosp

hate

,114

mg/

day

Cal

cium

,100

0 m

g/da

yIn

cide

nce

of v

erte

bral

es

tabl

ishe

d os

teop

oros

isfo

r 3

mon

ths

in 4

frac

ture

s(T

-sco

re <

–2.

5 an

d at

leas

t on

e G

roup

2:m

onof

luor

opho

spha

te,1

52 m

g/da

yos

teop

orot

ic v

erte

bral

frac

ture

)A

ll pa

tient

s re

ceiv

ed c

alci

um,1

000

mg/

day

Sebe

rt,1

99517

52

Vari

atio

ns in

lum

bar

BMD

Men

and

wom

en w

ith s

ever

e 60

.3 (

50–7

0)So

dium

mon

oflu

orop

hosp

hate

,100

mg/

day,

+

Plac

ebo

+ c

alci

um c

arbo

nate

,os

teop

aeni

a (lu

mba

r T-s

core

< –

2)

calc

ium

car

bona

te,2

500

mg/

day

(equ

ival

ent

to

2500

mg/

day

(equ

ival

ent

to

but

with

no

vert

ebra

l fra

ctur

esca

lciu

m,1

000

mg/

day)

1000

mg/

day

calc

ium

)

Appendix 5

236 Flu

ori

de:m

etho

dolo

gica

l qua

lity

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Broc

kste

dt,1

99673

11

11

01

5/15

(33

)92

41.3

Not

spe

cifie

d

Gut

teri

dge,

1993

801

11

10

15/

15 (

33)

Gro

up 1

:6N

ot s

peci

fied

Not

spe

cifie

dC

ontr

ol g

roup

1:8

Con

trol

gro

up 2

:9

Gut

teri

dge,

1996

791

11

11

16/

18 (

33)

6510

.8G

over

nmen

t/

rese

arch

cou

ncil

Han

sson

,198

7167

11

11

01

5/15

(33

)G

roup

1:2

512

Res

earc

h co

unci

l/G

roup

2:2

5ch

arita

ble

fund

ing

Gro

up 3

:25

Con

trol

gro

up:2

5

Kle

erek

oper

,1

13

21

19/

18 (

50)

Gro

up 1

:46

47.6

Gov

ernm

ent

1991

168

Con

trol

gro

up:3

6

Meu

nier

,199

8169

13

33

13

14/1

8 (7

8)G

roup

1:7

337

.3N

ot s

peci

fied

Gro

up 2

:68

Gro

up 3

:67

Con

trol

gro

up:1

46

Pak,

1995

170

13

23

13

13/1

8 (7

2)G

roup

1:5

430

.9G

over

nmen

t/C

ontr

ol g

roup

:56

inst

itutio

nal f

undi

ng

(pha

rmac

eutic

al c

ompa

ny,

drug

s on

ly)

Reg

inst

er,1

99817

13

33

31

316

/18

(89)

Gro

up 1

:100

39Ph

arm

aceu

tical

com

pany

Con

trol

gro

up:1

00

Rig

gs,1

99017

21

33

33

316

/18

(89)

Gro

up 1

:101

33.2

Gov

ernm

ent

body

Con

trol

gro

up:1

01

Rin

ge,1

99817

31

33

21

313

/18

(72)

Gro

up 1

:32

21.9

Not

spe

cifie

dC

ontr

ol g

roup

:32

Rin

ge,1

99917

41

33

21

313

/18

(72)

Gro

up 1

:49

27.6

Not

spe

cifie

dG

roup

2:4

8C

ontr

ol g

roup

:48

Sebe

rt,1

99517

51

13

33

314

/18

(78)

Gro

up 1

:45

37.2

Not

spe

cifie

dC

ontr

ol g

roup

:49


237


SE

RM

s:ge

nera

l inf

orm

atio

n

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Ettin

ger,

1999

182

3Pr

opor

tion

of w

omen

Po

stm

enop

ausa

l wom

en w

ith

67 (

31–8

0)G

roup

1:r

alox

ifene

,60

mg/

day

Plac

ebo

+ c

alci

um,5

00 m

g/da

y,+

w

ith o

ne o

r m

ore

new

os

teop

oros

is (

low

BM

D a

nd/o

r G

roup

2:r

alox

ifene

,120

mg/

day

vita

min

D3,

400–

600

IUno

n-tr

aum

atic

ver

tebr

al

vert

ebra

l fra

ctur

es);

37%

had

at

leas

t A

ll pa

tient

s re

ceiv

ed c

alci

um,5

00 m

g/da

y,+

fr

actu

res

one

vert

ebra

l fra

ctur

e at

ent

ryvi

tam

in D

3,40

0–60

0 IU

BMD

Lufk

in,1

99817

81

BMD

Post

men

opau

sal w

omen

with

68

.4G

roup

1:r

alox

ifene

,60

mg/

day

Plac

ebo

+ c

alci

um,7

50 m

g/da

y,+

Bi

oche

mic

al m

arke

rses

tabl

ishe

d os

teop

oros

is (

low

BM

D

Gro

up 2

:ral

oxife

ne,1

20 m

g/da

yvi

tam

in D

to

brin

g da

ily in

take

A

dver

se e

vent

san

d at

leas

t on

e no

n-tr

aum

atic

A

ll pa

tient

s re

ceiv

ed c

alci

um,7

50 m

g/da

y,+

to

800

IUve

rteb

ral f

ract

ure)

vita

min

D t

o br

ing

daily

inta

ke t

o 80

0 IU

SE

RM

s:m

etho

dolo

gica

l qua

lity

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Ettin

ger,

1999

182

33

33

13

16/1

8 (8

9)G

roup

1:2

557

23.4

Phar

mac

eutic

al c

ompa

nyG

roup

2:2

572

Con

trol

gro

up:2

576

Lufk

in,1

99817

81

13

21

311

/18

(61)

Gro

up 1

:48

7Ph

arm

aceu

tical

com

pany

Gro

up 2

:47

Con

trol

gro

up:4

8

Appendix 5

238 Pro

tein

sup

plem

ents

:gen

eral

info

rmat

ion

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Schu

rch,

1998

186

0.5

Insu

lin-li

ke g

row

th fa

ctor

-I El

derl

y m

en a

nd w

omen

with

rec

ent

80.7

Prot

ein,

20 g

,5 d

ays/

wee

k +

vita

min

D3,

Isoc

alor

ic p

lace

bo +

vita

min

D3,

leve

ls a

nd o

ther

os

teop

orot

ic h

ip fr

actu

re20

0,00

0 IU

,at

entr

y to

stu

dy20

0,00

0 IU

,at

entr

y to

stu

dybi

oche

mic

al d

ata

BMD

Pro

tein

sup

plem

ents

:met

hodo

logi

cal q

ualit

y

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Schu

rch,

1998

186

23

23

03

13/1

5 (8

7)G

roup

1:4

123

.2Ph

arm

aceu

tical

com

pany

/C

ontr

ol g

roup

:41

gove

rnm

ent

body


239


Vit

amin

K2:

gene

ral i

nfo

rmat

ion

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Shir

aki,

2000

682

Lum

bar

BMD

Am

bula

tory

wom

en w

ith o

steo

poro

sis

67.2

Vita

min

K2,

45 m

g/da

y,+

ele

men

tal c

alci

um,

Elem

enta

l cal

cium

,150

mg/

day

Vert

ebra

l fra

ctur

e(lu

mba

r BM

D <

70%

of y

oung

adu

lt15

0 m

g/da

ym

ean,

or w

ith o

ne o

r m

ore

non-

trau

mat

ic v

erte

bral

frac

ture

s an

d lu

mba

r BM

D <

80%

of y

oung

adu

lt m

ean)

Vit

amin

K2:

met

hodo

logi

cal q

ualit

y

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Shir

aki,

2000

681

31

31

312

/18

(67)

Gro

up 1

:120

21.2

Not

sta

ted

Con

trol

gro

up:1

21

Exe

rcis

e:ge

nera

l inf

orm

atio

n

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Ebra

him

,199

7187

2BM

DPo

stm

enop

ausa

l wom

en w

ith

68.0

Bris

k w

alki

ng,b

uild

ing

to 4

0 m

inut

es

Upp

er li

mb

exer

cise

ses

tabl

ishe

d os

teop

oros

is (

uppe

r th

ree

times

wee

kly

limb

frac

ture

in p

ast

2 ye

ars)

Exe

rcis

e:m

etho

dolo

gica

l qua

lity

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Ebra

him

,199

7187

23

22

12

12/1

8 (6

7)G

roup

1:8

141

Cha

rita

ble

trus

tC

ontr

ol g

roup

:84

Appendix 5

240 Co

mpa

riso

ns w

ith

acti

ve t

reat

men

ts:g

ener

al in

form

atio

n

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Abe

llan

Pere

z,1

Pain

Post

men

opau

sal w

omen

with

62

.4In

tran

asal

cal

cito

nin,

100

IU,+

ora

l ele

men

tal

Elem

enta

l cal

cium

,100

0 m

g/da

y19

9513

9N

ew v

erte

bral

frac

ture

ses

tabl

ishe

d os

teop

oros

is (

at le

ast

calc

ium

,500

mg/

day,

for

14 d

ays

out

of 2

8on

e no

n-tr

aum

atic

ver

tebr

al

crus

h fr

actu

re)

Ada

mi,

1995

932

BMD

(sp

ine)

Post

men

opau

sal w

omen

with

ost

eo-

59 (

48–7

6)G

roup

1:a

lend

rona

te,1

0 m

g/da

yIn

tran

asal

cal

cito

nin,

100

IU/d

ay,

poro

sis

(lum

bar T

-sco

re >

–2)

;5%

G

roup

2:a

lend

rona

te,2

0 m

g/da

y+

cal

cium

,500

mg/

day

had

vert

ebra

l fra

ctur

e at

ent

ryA

ll pa

tient

s re

ceiv

ed c

alci

um,5

00 m

g/da

y

Art

hur,

1990

190

1BM

DPo

stm

enop

ausa

l wom

en w

ith

66.4

Ora

l cal

citr

iol,

0.25

–0.5

0 µg

/day

,+ e

lem

enta

l El

emen

tal c

alci

um,1

g/da

y,ra

diog

raph

ic e

vide

nce

of o

steo

paen

ia;

calc

ium

,1g/

day

+ v

itam

in D

2,50

,000

U t

wic

e w

eekl

y40

% h

ad v

erte

bral

com

pres

sion

fr

actu

res

at e

ntry

Birk

enha

ger,

2BM

DPo

stm

enop

ausa

l wom

en w

ith61

.7 (

50–7

0)In

tram

uscu

lar

nand

rolo

ne d

ecan

oate

,50

mg

Oes

trad

iol v

aler

ate,

2 m

g/da

y,19

9216

4Bi

oche

mic

al m

arke

rses

tabl

ishe

d os

teop

oros

is (

at le

ast

one

ever

y 4

wee

ks,+

oes

trad

iol v

aler

ate,

2 m

g/da

y,da

ys 1

–25

of m

onth

+

non-

trau

mat

ic v

erte

bral

com

pres

sion

da

ys 1

–25

of m

onth

+ m

edro

xypr

oges

tero

ne

med

roxy

prog

este

rone

ace

tate

,fr

actu

re)

or o

steo

paen

ia (

lum

bar

acet

ate,

10 m

g/da

y,da

ys 1

6–25

of m

onth

10 m

g/da

y,da

ys 1

6–25

of m

onth

BMD

> 2

SD

bel

ow n

orm

al m

ean

of a

ge-m

atch

ed c

ontr

ols)

(8

6% h

ad fr

actu

res)

Ebel

ing,

1998

702

Vert

ebra

l fra

ctur

esM

en w

ith m

oder

atel

y se

vere

idio

path

icC

alci

trio

l,0.

50 µ

g/da

yC

alci

um,1

000

mg/

day

Spin

al B

MD

oste

opor

osis

(at

leas

t on

e ve

rteb

ral

frag

ility

frac

ture

)

Falc

h,19

8719

12

Bone

mas

s*Po

stm

enop

ausa

l wom

en w

ith

59.6

(50

–65)

Cal

citr

iol,

0.5

µg/d

ay (

redu

ced

to

Vita

min

D3,

400

IU/d

ayVe

rteb

ral f

ract

ures

*es

tabl

ishe

d os

teop

oros

is

0.25

µg/

day

in 2

8% o

f cas

es)

Frac

ture

s of

long

bon

es*

(frac

ture

of d

ista

l lef

t fo

rear

m)

Fujit

a,19

9264

Not

spe

cifie

dVe

rteb

ral f

ract

ures

Wom

en w

ith e

stab

lishe

d os

teop

oros

is

81In

tram

uscu

lar

calc

itoni

n,10

uni

ts,t

wic

e w

eekl

yA

lfaca

lcid

ol,0

.75–

1.5

µg/d

ay(a

t le

ast

one

non-

trau

mat

ic

vert

ebra

l fra

ctur

e)

Fujit

a,19

9358

48 w

eeks

Lum

bar

vert

ebra

l Pa

tient

s w

ith p

ostm

enop

ausa

l or

Not

spe

cifie

dG

roup

1:e

tidro

nate

,200

mg/

day,

for

2 w

eeks

of 1

2-w

eek

cycl

e +

pla

cebo

(al

faca

lcid

ol)

min

eral

den

sity

seni

le o

steo

poro

sis

Gro

up 2

:etid

rona

te,4

00 m

g/da

y,fo

r 2

wee

ks o

f 12-

wee

k cy

cle

+ p

lace

bo (

alfa

calc

idol

)G

roup

3:p

lace

bo (

etid

rona

te)

for

2 w

eeks

of 1

2-w

eek

cycl

e +

alfa

calc

idol

,1µg

/day

Gal

lagh

er,1

990b

743

Hyp

eros

teoi

dosi

s*M

en a

nd w

omen

with

est

ablis

hed

Not

spe

cifie

dFl

uori

de (

‘sta

ndar

d re

gim

e’)

+ c

alci

um,2

g/d

ayFl

uori

de (

‘sta

ndar

d re

gim

e’)

+

Tota

l bod

y ca

lciu

m*

idio

path

ic s

pina

l ost

eopo

rosi

sca

lcitr

iol,

0.5

µg/d

ayVe

rteb

ral f

ract

ures

*(a

t le

ast

one

vert

ebra

l fra

ctur

e)H

ip fr

actu

res*

Geu

sens

,198

6163

2Bo

ne m

iner

al c

onte

ntM

en a

nd w

omen

with

est

ablis

hed

Onl

y m

edia

n pe

r G

roup

1:i

ntra

mus

cula

r na

ndro

lone

dec

anoa

te,5

0 m

g,ev

ery

3 w

eeks

+ p

lace

bo

Cor

tical

bon

e vo

lum

eos

teop

oros

is (

at le

ast

one

grou

p gi

ven

(alfa

calc

idol

and

cal

cium

infu

sion

s)Ve

rteb

ral f

ract

ure

rate

vert

ebra

l fra

ctur

e)R

x2:a

lfaca

lcid

ol,1

µg/d

ay,+

pla

cebo

(na

ndro

lone

and

cal

cium

infu

sion

s)Bi

oche

mic

al v

aria

bles

Rx3

:int

rave

nous

cal

cium

infu

sion

s,15

mg/

kg b

ody

wei

ght,

for

12 c

onse

cutiv

e da

ys

annu

ally

+ p

lace

bo (

nand

rolo

ne a

nd a

lfaca

lcid

ol)

cont

inue

d


241


Co

mpa

riso

ns w

ith

acti

ve t

reat

men

ts:g

ener

al in

form

atio

n co

ntd

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Gua

ñabe

ns,1

99675

2Bo

ne m

ass

Wom

en w

ith e

stab

lishe

d64

.5O

ral f

luor

ide,

50 m

g/da

y,+

cal

cium

,1g,

‘reg

ular

ly’

Etid

rona

te,4

00 m

g/da

y,fo

r 14

day

s Ve

rteb

ral f

ract

ure

post

men

opau

sal o

steo

poro

sis

(mea

n ou

t of

74

+ c

alci

um,1

g ‘r

egul

arly

’lu

mba

r BM

D T

-sco

re –

3.2

SD p

lus

at le

ast

one

atra

umat

ic v

erte

bral

fr

actu

re)

Gut

teri

dge,

1996

792.

25BM

D*

Post

men

opau

sal w

omen

with

69.6

HRT

+ e

nter

ic-c

oate

d so

dium

fluo

ride

,60

mg/

day,

HRT

+ c

alci

um,1

g/da

yVe

rteb

ral f

ract

ures

*es

tabl

ishe

d os

teop

oros

is (

at le

ast

for

mon

ths

1–6

of 9

-mon

th c

ycle

+ c

alci

um,1

g/da

yon

e ve

rteb

ral f

ract

ure)

Leid

ig-B

ruck

ner,

4.7

±1.

6BM

D*

Post

men

opau

sal w

omen

with

64.4

Gro

up 1

:sod

ium

fluo

ride

,75

mg/

day

1997

76Ve

rteb

ral f

ract

ures

*es

tabl

ishe

d os

teop

oros

is (

at le

ast

Gro

up 2

:sod

ium

fluo

ride

,75

mg/

day,

+ o

estr

adio

l val

erat

e,1

mg/

day,

or o

estr

adio

lon

e ve

rteb

ral f

ract

ure)

vale

rate

,1.2

5 m

g/da

y,+

med

roxy

prog

este

rone

ace

tate

,5 m

g/da

yG

roup

3:s

odiu

m fl

uori

de,7

5 m

g/da

y,+

intr

amus

cula

r na

ndro

lone

dec

anoa

te,5

0 m

g,ev

ery

4 w

eeks

for

2 ye

ars

All

patie

nts

rece

ived

cal

cium

,100

0 m

g,+

vita

min

D,3

000

IU,d

aily

Lem

s,19

9719

22

BMD

Men

and

wom

en w

ith e

stab

lishe

d 58

Etid

rona

te,4

00 m

g/da

y,fo

r 2

wee

ks o

f 13-

wee

kEt

idro

nate

,400

mg/

day,

for

2 w

eeks

of

cort

icos

tero

id-in

duce

d os

teop

oros

is

cycl

e +

sod

ium

fluo

ride

,50

mg/

day,

+ c

alci

um,

13-w

eek

cycl

e +

pla

cebo

(flu

orid

e +

(v

erte

bral

def

orm

ity,p

rior

50

–100

mg/

day

calc

ium

,50–

100

mg/

day)

peri

pher

al fr

actu

re o

r bo

th)

Lyri

tis,1

99416

61

Vert

ebra

l fra

ctur

esPo

stm

enop

ausa

l wom

en w

ith

66.9

Intr

amus

cula

r na

ndro

lone

dec

anoa

te,5

0 m

g,A

lfaca

lcid

ol,1

µg/d

ay,+

pla

cebo

BM

Des

tabl

ishe

d os

teop

oros

is (

at le

ast

ever

y 3

wee

ks +

pla

cebo

(al

faca

lcid

ol)

(nan

drol

one)

Sym

ptom

s (p

ain,

mob

ility

)on

e no

n-tr

aum

atic

ver

tebr

al c

olla

pse)

Mam

elle

,198

8176

2Ve

rteb

ral f

ract

ures

Men

and

wom

en w

ith e

stab

lishe

d 70

.9 (

50–9

2)So

dium

fluo

ride

,50

mg/

day,

elem

enta

l cal

cium

,O

ne o

f oth

er r

egim

es p

resc

ribe

d by

os

teop

oros

is (

at le

ast

one

non-

1g/

day,

+ v

itam

in D

2,80

0 IU

/day

Fren

ch p

hysi

cian

s (c

alci

um +

vita

min

tr

aum

atic

ver

tebr

al c

rush

frac

ture

)D

2,n

= 9

5;cy

clic

al c

alci

toni

n +

ph

osph

orus

,n=

85;

calc

itoni

n +

ca

lciu

m,n

= 1

2;ca

lciu

m +

ph

osph

orus

,n=

17;

phos

phor

us +

cy

clic

al e

tidro

nate

,n=

2)

Paci

fici,

1988

104

2Bo

ne m

iner

al c

onte

ntW

omen

with

ost

eopo

rosi

s or

58

(26

–80)

Pota

ssiu

m p

hosp

hate

,150

0 m

g/da

y,on

day

s 1–

3 C

onju

gate

d oe

stro

gens

,0.6

25 m

g/da

y,os

teop

aeni

a (a

t le

ast

one

non-

and

etid

rona

te,4

00 m

g/da

y,on

day

s 4–

17 o

f 73-

day

for

25 d

ays/

mon

th,

trau

mat

ic v

erte

bral

frac

ture

and

/or

cycl

e +

cal

cium

,100

0 m

g/da

ym

edro

xypr

oges

tero

ne,1

0 m

g/da

y,da

ys

evid

ence

of s

pina

l dem

iner

alis

atio

n)15

–25

+ c

alci

um,1

000

mg/

day

Pak,

1989

193

2.9

(mea

n)BM

D*

Men

and

wom

en w

ith o

steo

poro

sis

62.4

(29

–82)

Wee

ks 1

–2 o

f 5-m

onth

cyc

le:1

,25-

hydr

oxyv

itam

inW

eeks

3–1

4 of

5-m

onth

cyc

le:

Vert

ebra

l fra

ctur

es*

or o

steo

paen

ia (

frac

ture

s,lo

w B

MD

D,2

mg/

day;

wee

ks 3

–14:

sodi

um fl

uori

de,5

0 m

g/da

y,so

dium

fluo

ride

,50

mg/

day,

+ 2

5-N

on-v

erte

bral

frac

ture

s*or

rad

iolo

gica

l evi

denc

e of

+ 2

5-hy

drox

yvita

min

D,5

0 µg

,tw

ice

wee

kly

+hy

drox

yvita

min

D,5

0 µg

,tw

ice

Bone

bio

psy*

oste

opae

nia)

(98

% h

ad v

erte

bral

ca

lciu

m s

uppl

emen

ts t

o br

ing

daily

inta

ke t

ow

eekl

y +

cal

cium

sup

plem

ents

to

or p

erip

hera

l fra

ctur

es)

1500

mg/

day;

wee

ks 1

5–20

:25-

hydr

oxyv

itam

in D

,br

ing

daily

inta

ke t

o 15

00 m

g/da

y;50

µg,

twic

e w

eekl

y +

cal

cium

sup

plem

ents

to

wee

ks 1

5–20

:25-

hydr

oxyv

itam

in D

,br

ing

daily

inta

ke t

o 15

00 m

g/da

y50

µg,

twic

e w

eekl

y +

cal

cium

su

pple

men

ts t

o br

ing

daily

inta

ke

to 1

500

mg/

day

cont

inue

d

Appendix 5

242 Co

mpa

riso

ns w

ith

acti

ve t

reat

men

ts:g

ener

al in

form

atio

n co

ntd

Stu

dyL

engt

h o

f P

rim

ary

out

com

ePo

pula

tio

nM

ean

age

Inte

rven

tio

n/do

seC

om

pari

son/

sst

udy

(yea

rs)

mea

sure

/s(r

ange

) (y

ears

)

Roz

hins

kaya

,1

BMD

Wom

en w

ith e

stab

lishe

d st

eroi

d-56

.3G

roup

1:f

luor

ide,

15 m

g/da

y,+

cal

cium

,450

mg/

day

1999

72Ve

rteb

ral f

ract

ures

indu

ced

oste

opor

osis

(at

leas

t tw

o G

roup

2:f

luor

ide,

15 m

g/da

y,+

alfa

calc

idol

,0.5

µg/

day,

+ c

alci

um,4

50 m

g/da

yBa

ck p

ain

vert

ebra

l fra

ctur

es +

T-s

core

< –

2.5

Gro

up 3

:alfa

calc

idol

,0.7

5 µg

/day

at s

pine

or

fem

ur n

eck)

Shio

ta,1

99819

42

BMD

Wom

en a

ged

over

50

year

s w

ith

68.1

Gro

up 1

:ipr

iflav

one,

600

mg/

day

post

men

opau

sal o

r se

nile

ost

eo-

Gro

up 2

:sub

cuta

neou

s ca

lcito

nin,

20 IU

/wee

kpo

rosi

s (B

MD

less

tha

n 0.

70 g

/cm

2 )G

roup

3:a

lfaca

lcid

ol,0

.5 µ

g/da

y,+

calc

ium

lact

ate,

2 g/

day

Gro

up 4

:alfa

calc

idol

,2 µ

g/da

y,fo

r w

eeks

1–2

,and

0.5

µg/

day

for

wee

ks 7

–14;

ipri

flavo

ne,6

00 m

g/da

y,fo

r w

eeks

3–6

;cal

cium

lact

ate,

2 g/

day,

for

wee

ks 7

–14

of 1

4–w

eek

cycl

e

Shir

aki,

1999

195

48 w

eeks

Lum

bar

BMD

Wom

en w

ith p

rim

ary

oste

opor

osis

63

.3O

ral a

lend

rona

te,5

mg/

day

+ e

lem

enta

l A

lfaca

lcid

ol,1

µg/d

ay,+

ele

men

tal

(lum

bar T

-sco

re <

–2.

5,or

< –

1.5

with

ca

lciu

m,2

00 m

g/da

yca

lciu

m,2

00 m

g/da

yon

e or

mor

e ve

rteb

ral f

ract

ures

)

Thi

ébau

d,19

9411

52

BMD

Post

men

opau

sal w

omen

with

65

.9In

trav

enou

s pa

mid

rona

te,3

00 m

g/m

onth

,Fl

uori

de,2

0–30

mg/

day,

+ c

alci

um,

esta

blis

hed

oste

opor

osis

(at

leas

t +

cal

cium

,1g/

day,

and

vita

min

D,1

000

U/d

ay1

g/da

y,an

d vi

tam

in D

,100

0 U

/day

one

vert

ebra

l fra

ctur

e)

Tily

ard,

1992

124

3R

ate

of n

ew v

erte

bral

Po

stm

enop

ausa

l wom

en w

ith63

.7 (

50–7

9)O

ral c

alci

trio

l,0.

25 µ

g/da

yEl

emen

tal c

alci

um,1

g/da

yfr

actu

res

esta

blis

hed

oste

opor

osis

(at

leas

t on

e no

n-tr

aum

atic

ver

tebr

alco

mpr

essi

on fr

actu

re)

Wat

ts,1

99010

92

(as

RC

T)

Spin

al B

MD

Post

men

opau

sal w

omen

with

65.1

Gro

up 1

:sod

ium

-pot

assi

um p

hosp

hate

,2 g

/day

,on

days

1–3

+ p

lace

bo (

etid

rona

te)

on

Rat

e of

new

ver

tebr

al

esta

blis

hed

oste

opor

osis

(at

leas

tda

ys 4

–17

of 9

1-da

y cy

cle

frac

ture

son

e bu

t no

mor

e th

an fo

ur

Gro

up 2

:pla

cebo

(so

dium

-pot

assi

um p

hosp

hate

) on

day

s 1–

3 +

etid

rona

te,4

00 m

g/da

y,ve

rteb

ral c

rush

frac

ture

s)on

day

s 4–

17 o

f 91-

day

cycl

eG

roup

3:s

odiu

m-p

otas

sium

pho

spha

te,2

g/d

ay,o

n da

ys 1

–3,e

tidro

nate

,400

mg/

day,

on

days

4–1

7 of

91-

day

cycl

eA

ll pa

tient

s re

ceiv

ed c

alci

um,5

00 m

g/da

y,on

day

s 18

-91

of 9

1-da

y cy

cle

Wim

alaw

ansa

,4

BMD

Post

men

opau

sal w

omen

with

64.9

(58

–72)

Gro

up 1

:Pre

mar

in®,0

.625

mg/

day,

+ N

orge

stre

l®,1

50 µ

g,fo

r 12

day

s/m

onth

1998

110

esta

blis

hed

oste

opor

osis

(sp

inal

G

roup

2:e

tidro

nate

,400

mg/

day,

for

wee

ks 1

–2 o

f 12-

wee

k cy

cle

T-sc

ore

–2 a

nd a

t le

ast

one

but

no

Gro

up 3

:Pre

mar

in®,0

.625

mg/

day,

+ N

orge

stre

l®,1

50 µ

g,fo

r 12

day

s/m

onth

+ e

tidro

nate

,m

ore

than

four

atr

aum

atic

tho

raci

c 40

0 m

g/da

y,fo

r w

eeks

1–2

of 1

2-w

eek

cycl

eve

rteb

ral c

rush

frac

ture

s)A

ll pa

tient

s re

ceiv

ed e

lem

enta

l cal

cium

,1g/

day,

and

vita

min

D,4

00 U

/day


243


Co

mpa

riso

ns w

ith

acti

ve t

reat

men

ts:m

etho

dolo

gica

l qua

lity

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Abe

llan

Pere

z,2

12

30

310

/15

(67)

Gro

up 1

:43

Not

spe

cifie

dN

ot s

peci

fied

1995

139

Con

trol

gro

up:4

5

Ada

mi,

1995

931

13

31

09/

15 (

73)

Gro

up 1

:68

14.3

Not

spe

cifie

dG

roup

2:7

2G

roup

3:7

5

Art

hur,

1990

190

13

11

01

7/15

(47

)14

28.6

Not

spe

cifie

d

Birk

enha

ger,

13

21

03

10/1

5 (6

7)43

16.3

Phar

mac

eutic

al c

ompa

ny19

9216

4

Ebel

ing,

1998

701

31

11

18/

18 (

44)

Gro

up 1

:20

12%

Not

spe

cifie

dG

roup

2:1

9

Falc

h,19

8719

11

32

31

313

/18

(72)

Gro

up 1

:47

11.6

Not

spe

cifie

dC

ontr

ol g

roup

:39

Fujit

a,19

9264

13

22

03

11/1

5 (7

3)G

roup

1:8

6.2

Not

spe

cifie

dG

roup

2:8

Con

trol

gro

up:8

Fujit

a,19

9358

11

12

01

6/15

(40

)41

4N

ot s

peci

fied

Not

spe

cifie

d

Gal

lagh

er,1

990b

741

11

11

16/

18 (

33)

40N

ot s

peci

fied

Not

spe

cifie

d

Geu

sens

,198

6163

13

21

03

10/1

5 (6

7)60

56.7

Phar

mac

eutic

al c

ompa

ny

(dru

gs o

nly)

Gua

ñabe

ns,1

99675

11

21

11

7/18

(39

)G

roup

1:6

028

.8N

ot s

peci

fied

Gro

up 2

:65

Gut

teri

dge,

1996

791

11

11

16/

18 (

33)

Gro

up 1

:14

14.3

Gov

ernm

ent/

Gro

up 2

:16

rese

arch

cou

ncil

Leid

ig-B

ruck

ner,

11

13

01

7/15

(47

)G

roup

1:1

77.

9N

ot s

peci

fied

1997

76G

roup

2:8

Gro

up 3

:13

Lem

s,19

9719

21

12

13

311

/18

(61)

Gro

up 1

:24

14.9

Cha

rita

ble

foun

datio

n/G

roup

2:2

3ph

arm

aceu

tical

com

pany

Lyri

tis,1

99416

61

13

30

19/

15 (

60)

Gro

up 1

:44

0N

ot s

peci

fied

Gro

up 2

:44

cont

inue

d

Appendix 5

244 Co

mpa

riso

ns w

ith

acti

ve t

reat

men

ts:m

etho

dolo

gica

l qua

lity

cont

d

Stu

dyR

ando

mis

atio

nB

lindi

ng o

f H

andl

ing

of

Co

mpa

rabi

lity

Dia

gno

sis

of

Dia

gno

sis

Tota

lN

umbe

r o

f pa

tien

tsL

ost

to

S

our

ce o

f fu

ndin

gfr

actu

re o

utco

me

wit

hdra

wal

so

f gr

oup

sno

n-ve

rteb

ral

of

vert

ebra

lm

etho

dolo

gyra

ndo

mis

ed

follo

w-u

p as

sess

ors

at e

ntry

frac

ture

frac

ture

sco

re (

%)

to s

tudy

(%)

Mam

elle

,198

8176

11

32

13

11/1

8 (6

1)G

roup

1:2

5713

Phar

mac

eutic

al c

ompa

ny/

Gro

up 2

:209

gove

rnm

ent

agen

cy

Paci

fici,

1988

104

11

23

03

10/1

5 (6

7)12

845

.3N

ot s

peci

fied

Pak,

1989

193

13

32

33

15/1

8 (8

3)G

roup

1:2

3N

ot s

peci

fied

Gro

up 2

:21

Roz

hins

kaya

,1

12

10

16/

15 (

40)

Gro

up 1

:6N

ot s

peci

fied

1999

72G

roup

2:6

Gro

up 3

:10

Shio

ta,1

99819

41

11

10

26/

15 (

40)

Not

cle

arN

ot c

lear

Not

spe

cifie

d

Shir

aki,

1999

195

31

13

03

11/1

5 (7

3)G

roup

1:1

0540

.5Ph

arm

aceu

tical

com

pany

Gro

up 2

:105

Thi

ébau

d,19

9411

51

12

31

210

/18

(56)

Gro

up 1

:16

6.2

Dru

g do

nate

d by

Con

trol

gro

up:1

6ph

arm

aceu

tical

com

pany

Tily

ard,

1992

124

23

33

32

16/1

8 (8

9)G

roup

1:3

1430

.6N

ot s

peci

fied

Con

trol

gro

up:3

08

Wat

ts,1

99010

92

33

31

315

/18

(83)

Gro

up 1

:107

15.4

Phar

mac

eutic

al c

ompa

nyG

roup

2:1

05G

roup

3:1

07

Wim

alaw

ansa

,2

32

31

314

/18

(78)

Gro

up 1

:18

19.4

No

info

rmat

ion

1998

110

Gro

up 2

:17

Gro

up 3

:19


245

1 Economics/2 exp “Costs and cost analysis”/3 Economic value of life/4 exp Economics, hospital/5 exp Economics, medical/6 Economics, nursing/7 exp models, economic/8 Economics, pharmaceutical/9 exp “Fees and charges”/10 exp Budgets/11 ec.fs.12 (cost or costs or costed or costly or

costing$).tw.13 (economic$ or pharmacoeconomic$ or price$

or pricing).tw.14 or/1–1315 exp Osteoporosis/16 Bone diseases, metabolic/17 osteoporo$.tw. 18 15 or 16 or 1719 (bone adj6 densit$).tw.20 Bone density/21 (bone or bones).mp.22 exp Densitometry/23 Tomography, x-ray computed/24 densit$.tw.25 23 and 2426 22 or 2527 21 and 2628 19 or 20 or 2729 Colles’ fracture/30 exp Hip fractures/31 Spinal fractures/32 29 or 30 or 31

33 Fractures/34 colles$.tw.35 9hip or hips).tw.36 (femur adj6 neck).tw.37 (femoral adj6 neck).tw.38 (spine or spinal).tw.39 vertebra$.tw.40 Lumbar vertebrae/41 34 or 35 or 36 or 37 or 38 or 39 or 4042 33 and 4143 fractur$.tw.44 34 or 35 or 36 or 37 or 38 or 3945 (fractur$ adj6 (colles$ or (hip or hips) or

(femur adj6 neck) or (femoral adj6 neck) or(spine or spinal) or vertebra$)).tw.

46 32 or 42 or 4547 Estrogen replacement therapy/48 estrogen replacement therapy.tw.49 oestrogen replacement therapy.tw.50 hormone replacement therapy.tw.51 ert.tw.52 ort.tw.53 hrt.tw.54 47 or 48 or 49 or 50 or 51 or 52 or 5355 exp Menopause/56 Climacteric/57 menopaus$.tw.58 postmenopaus$.tw.59 climacteric.tw.60 55 or 56 or 57 or 58 or 5961 54 or 6062 46 and 6163 18 or 28 or 6264 14 and 63


Appendix 6

Economics literature search strategy


253


Health Technology Assessment Programme

continued

Prioritisation Strategy Group

Chair,Professor Kent Woods,Director, NHS HTA Programme, &Professor of TherapeuticsUniversity of Leicester

Professor Bruce Campbell,Consultant Vascular & General Surgeon, Royal Devon & Exeter Hospital

Professor Shah Ebrahim,Professor in Epidemiology of Ageing, University of Bristol

Dr John Reynolds, Clinical Director, Acute General Medicine SDU,Oxford Radcliffe Hospital

Dr Ron Zimmern, Director, Public HealthGenetics Unit, Strangeways ResearchLaboratories, Cambridge

Members

Current and past membership details of all HTA ‘committees’ are available from the HTA website (see inside front cover for details)

HTA Commissioning Board

Programme Director,Professor Kent Woods,Director, NHS HTA Programme, &Professor of TherapeuticsUniversity of Leicester

Chair,Professor Shah Ebrahim,Professor in Epidemiology of Ageing,University of Bristol

Deputy Chair,Professor Jon Nicholl,Director, Medical CareResearch Unit,University of Sheffield

Professor Douglas Altman,Director, ICRF MedicalStatistics Group, University of Oxford

Professor John Bond, Director, Centre for HealthServices Research, University of Newcastle-upon-Tyne

Professor John Brazier,Director of Health Economics,University of Sheffield

Dr Andrew Briggs, Research Fellow, Institute of Health Sciences,University of Oxford

Ms Christine Clark, Freelance Medical Writer, Bury, Lancs

Professor Martin Eccles,Professor of Clinical Effectiveness, University of Newcastle-upon-Tyne

Dr Andrew Farmer, General Practitioner & NHS R&D Clinical Scientist,Institute of Health Sciences,University of Oxford

Professor Adrian Grant,Director, Health ServicesResearch Unit, University of Aberdeen

Dr Alastair Gray, Director, Health EconomicsResearch Centre, Institute of Health Sciences, University of Oxford

Professor Mark Haggard,Director, MRC Institute of Hearing Research, University of Nottingham

Professor Jenny Hewison,Academic Unit of Psychiatry & Behavioural Sciences,University of Leeds

Professor Peter Jones,University Department of Psychiatry, University of Cambridge

Professor Alison Kitson,Director, Royal College ofNursing Institute, London

Professor Sarah Lamb,Research Professor in Physiotherapy, University of Coventry

Dr Donna Lamping, Head, Health ServicesResearch Unit, London School of Hygiene & Tropical Medicine

Professor David Neal,Department of Surgery,University of Newcastle-upon-Tyne

Professor Tim Peters, Social Medicine, University of Bristol

Professor Martin Severs,Professor in Elderly Health Care, University of Portsmouth

Dr Jonathan Shapiro, Senior Fellow, Health ServicesManagement Centre,Birmingham

Dr Sarah Stewart-Brown,Director, Health ServicesResearch Unit, University of Oxford

Dr Gillian Vivian, Consultantin Nuclear Medicine &Radiology, Royal CornwallHospitals Trust, Truro

Members


254

continued

Diagnostic Technologies & Screening Panel

Chair,Dr Ron Zimmern,Director, Public HealthGenetics Unit, StrangewaysResearch Laboratories,Cambridge

Mrs Stella Burnside, Chief Executive, Altnagelvin Hospitals Health & Social Services Trust,Londonderry

Dr Paul O Collinson,Consultant ChemicalPathologist & Senior Lecturer, St George's Hospital, London

Dr Barry Cookson, Director, Laboratory ofHospital Infection, PublicHealth Laboratory Service,London

Professor Howard Cuckle,Professor of ReproductiveEpidemiology, University of Leeds

Professor Adrian K Dixon,Professor of Radiology,Addenbrooke's Hospital,Cambridge

Dr David Elliman, Consultant in CommunityChild Health, St. George’s Hospital, London

Dr Tom Fahey, Senior Lecturer in General Practice, University of Bristol

Dr Andrew Farmer, General Practitioner & NHSR&D Clinical Scientist,Institute of Health Sciences,University of Oxford

Professor Jane Franklyn,Professor of Medicine,University of Birmingham

Dr Antony J Franks, Deputy Medical Director, The Leeds Teaching HospitalsNHS Trust

Dr J A Muir Gray, Programmes Director, National ScreeningCommittee, NHS Executive, Oxford

Dr Peter Howlett, Executive Director – Planning,Portsmouth Hospitals NHS Trust

Dr S M Ludgate, Medical Director, Medical Devices Agency,London

Professor Jennie Popay,Professor of Sociology & Public Health, Institute for Health Research,University of Lancaster

Dr Susan Schonfield, CPHM SpecialistCommissioning, Public Health Directorate,Croydon Primary Care Trust

Mrs Kathlyn Slack, Professional Support,Diagnostic Imaging &Radiation Protection Team,Department of Health, London

Mr Tony Tester, Chief Officer, South BedfordshireCommunity Health Council,Luton

Dr Andrew Walker, Senior Lecturer in Health Economics, University of Glasgow

Professor Martin J Whittle,Head of Division ofReproductive & Child Health,University of Birmingham

Members


Pharmaceuticals Panel

Chair,Dr John Reynolds,Clinical Director, Acute General Medicine SDU,Oxford Radcliffe Hospital

Professor Tony Avery, Professor of Primary Health Care, University of Nottingham

Professor Iain T Cameron,Professor of Obstetrics & Gynaecology, University of Southampton

Mr Peter Cardy, Chief Executive, Macmillan Cancer Relief,London

Dr Christopher Cates, GP & Cochrane Editor, Bushey Health Centre, Bushey, Herts

Dr Karen A Fitzgerald,Pharmaceutical Adviser, Bro Taf Health Authority,Cardiff

Dr Felicity J Gabbay, Managing Director, Transcrip Ltd, Milford-on-Sea, Hants

Mr Peter Golightly, Director, Trent MedicinesInformation Services, Leicester Royal Infirmary

Dr Alastair Gray, Director, Health EconomicsResearch Centre, Institute of Health Sciences, University of Oxford

Mrs Sharon Hart, Managing Editor, Drug &Therapeutics Bulletin, London

Dr Christine Hine, Consultant in Public Health Medicine, Bristol South & West Primary Care Trust

Mrs Jeannette Howe, Deputy Chief Pharmacist,Department of Health, London

Professor Robert Peveler,Professor of Liaison Psychiatry,Royal South Hants Hospital,Southampton

Dr Frances Rotblat, CPMP Delegate, Medicines Control Agency,London

Dr Eamonn Sheridan,Consultant in ClinicalGenetics, St James's University Hospital, Leeds

Mrs Katrina Simister, New Products Manager,National Prescribing Centre,Liverpool

Professor Terence Stephenson,Professor of Child Health,University of Nottingham

Dr Richard Tiner, Medical Director, Association of the BritishPharmaceutical Industry,London

Professor Jenifer Wilson-Barnett, Head of FlorenceNightingale School of Nursing& Midwifery, King's College,London

Members


255



Therapeutic Procedures Panel

Chair,Professor Bruce Campbell,Consultant Vascular & General Surgeon, Royal Devon& Exeter Hospital

Professor John Bond, Professor of Health Services Research, Centre forHealth Services Research,University of Newcastle-upon-Tyne

Ms Judith Brodie, Head of Cancer Support Service, Cancer BACUP, London

Ms Tracy Bury, Head of Research & Development, Chartered Society of Physiotherapy, London

Mr Michael Clancy, Consultant in A & E Medicine,Southampton General Hospital

Professor Collette Clifford,Professor of Nursing & Head of Research, School of Health Sciences,University of Birmingham

Dr Carl E Counsell, Senior Lecturer in Neurology,University of Aberdeen

Dr Keith Dodd, Consultant Paediatrician,Derbyshire Children's Hospital,Derby

Mr John Dunning, Consultant CardiothoracicSurgeon, Papworth HospitalNHS Trust, Cambridge

Mr Jonothan Earnshaw,Consultant Vascular Surgeon,Gloucestershire Royal Hospital,Gloucester

Professor Gene Feder,Professor of Primary Care R&D, St Bartholomew’s & the London, Queen Mary's School ofMedicine & Dentistry, University of London

Professor Richard Johanson,Consultant & Senior Lecturer,North Staffordshire InfirmaryNHS Trust, Stoke-on-Trent(deceased Feb 2002)

Dr Duncan Keeley, General Practitioner, Thame, Oxon

Dr Phillip Leech, Principal Medical Officer for Primary Care, Department of Health, London

Mr George Levvy, Chief Executive, Motor Neurone Disease Association,Northampton

Professor James Lindesay,Professor of Psychiatry for the Elderly, University of Leicester

Professor Rajan Madhok,Medical Director & Director of Public Health, North & East Yorkshire & Northern LincolnshireStrategic Health Authority,York

Dr Mike McGovern, Senior Medical Officer, Heart Team, Department of Health, London

Dr John C Pounsford,Consultant Physician, Frenchay Healthcare Trust,Bristol

Professor Mark Sculpher,Professor of Health Economics, Institute for Research in the Social Services,University of York

Dr Ken Stein, Senior Lecturer in Public Health, Peninsular TechnologyAssessment Group, University of Exeter

Members

continued


256

continued


Expert Advisory Network

Mr Gordon Aylward,Chief Executive,Association of British Health-Care Industries,London

Mr Shaun Brogan,Chief Executive,Ridgeway Primary Care Group,Aylesbury, Bucks

Mr John A Cairns,Reader in Health Economics,Health Economics Research Unit,University of Aberdeen

Professor Nicky Cullum,Director of Centre forEvidence-Based Nursing,University of York

Dr Katherine Darton,Information Unit,MIND -- The Mental Health Charity, London

Professor Carol Dezateux,Professor of Paediatric Epidemiology,Institute of Child Health,London

Professor Pam Enderby,Dean of Faculty of MedicineInstitute of General Practice & Primary Care,University of Sheffield

Mr Leonard R Fenwick,Chief Executive, Freeman Hospital,Newcastle-upon-Tyne

Professor David Field,Professor of Neonatal Medicine,The Leicester Royal Infirmary NHS Trust

Mrs Gillian Fletcher,Antenatal Teacher & Tutor & President,National Childbirth Trust, Henfield, West Sussex

Ms Grace Gibbs,Deputy Chief ExecutiveDirector for Nursing,Midwifery & Clinical Support Services,West Middlesex University Hospital,Isleworth, Middlesex

Dr Neville Goodman,Consultant Anaesthetist,Southmead Hospital, Bristol

Professor Robert E Hawkins,CRC Professor & Director of Medical Oncology,Christie Hospital NHS Trust,Manchester

Professor F D Richard Hobbs,Professor of Primary Care & General Practice,University of Birmingham

Professor Allen Hutchinson,Director of Public Health &Deputy Dean of ScHARR,University of Sheffield

Professor David Mant,Professor of General Practice,Institute of Health Sciences,University of Oxford

Professor Alexander Markham,Director, Molecular Medicine Unit,St James's University Hospital,Leeds

Dr Chris McCall,General Practitioner,The Hadleigh Practice,Corfe Mullen, Dorset

Professor Alistair McGuire,Professor of Health Economics,London School of Economics,University of London

Dr Peter Moore,Freelance Science Writer,Ashtead, Surrey

Dr Andrew Mortimore,Consultant in Public Health Medicine,Southampton City PrimaryCare Trust

Dr Sue Moss,Associate Director,Cancer Screening Evaluation Unit,Institute of Cancer Research,Sutton, Surrey

Mrs Julietta Patnick,National Coordinator,NHS Cancer Screening Programmes,Sheffield

Professor Chris Price,Director of Clinical Research,Bayer Diagnostics Europe,Stoke Poges, Berks

Ms Marianne Rigge,Director, College of Health,London

Dr William Rosenberg,Senior Lecturer & Consultant in Medicine,University of Southampton

Professor Ala Szczepura,Director, Centre for Health Services Studies,University of Warwick

Dr Ross Taylor,Senior Lecturer,Department of GeneralPractice & Primary Care,University of Aberdeen

Mrs Joan Webster,Consumer member,HTA -- Expert Advisory Network

Members

Health Technology Assessm

ent 2002;Vol.6:No.29

Treatment of established osteoporosis

Copies of this report can be obtained from:

The National Coordinating Centre for Health Technology Assessment,Mailpoint 728, Boldrewood,University of Southampton,Southampton, SO16 7PX, UK.Fax: +44 (0) 23 8059 5639 Email: [email protected]://www.ncchta.org ISSN 1366-5278

FeedbackThe HTA Programme and the authors would like to know

your views about this report.

The Correspondence Page on the HTA website(http://www.ncchta.org) is a convenient way to publish

your comments. If you prefer, you can send your comments to the address below, telling us whether you would like

us to transfer them to the website.

We look forward to hearing from you.

ja kanis je brazier m stevenson nw...

Documents